Next Article in Journal
Simple and Low-Cost Photovoltaic Module Emulator
Previous Article in Journal
The Technical Challenges Facing the Integration of Small-Scale and Large-scale PV Systems into the Grid: A Critical Review
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Demosaicing of Bayer and CFA 2.0 Patterns for Low Lighting Images

Applied Research LLC, Rockville, MD 20850, USA
*
Author to whom correspondence should be addressed.
Electronics 2019, 8(12), 1444; https://doi.org/10.3390/electronics8121444
Submission received: 26 October 2019 / Revised: 24 November 2019 / Accepted: 26 November 2019 / Published: 1 December 2019
(This article belongs to the Section Circuit and Signal Processing)

Abstract

:
It is commonly believed that having more white pixels in a color filter array (CFA) will help the demosaicing performance for images collected in low lighting conditions. However, to the best of our knowledge, a systematic study to demonstrate the above statement does not exist. We present a comparative study to systematically and thoroughly evaluate the performance of demosaicing for low lighting images using two CFAs: the standard Bayer pattern (aka CFA 1.0) and the Kodak CFA 2.0 (RGBW pattern with 50% white pixels). Using the clean Kodak dataset containing 12 images, we first emulated low lighting images by injecting Poisson noise at two signal-to-noise (SNR) levels: 10 dBs and 20 dBs. We then created CFA 1.0 and CFA 2.0 images for the noisy images. After that, we applied more than 15 conventional and deep learning based demosaicing algorithms to demosaic the CFA patterns. Using both objectives with five performance metrics and subjective visualization, we observe that having more white pixels indeed helps the demosaicing performance in low lighting conditions. This thorough comparative study is our first contribution. With denoising, we observed that the demosaicing performance of both CFAs has been improved by several dBs. This can be considered as our second contribution. Moreover, we noticed that denoising before demosaicing is more effective than denoising after demosaicing. Answering the question of where denoising should be applied is our third contribution. We also noticed that denoising plays a slightly more important role in 10 dBs signal-to-noise ratio (SNR) as compared to 20 dBs SNR. Some discussions on the following phenomena are also included: (1) why CFA 2.0 performed better than CFA 1.0; (2) why denoising was more effective before demosaicing than after demosaicing; and (3) why denoising helped more at low SNRs than at high SNRs.

1. Introduction

The standard Bayer pattern [1], also known as color filter array (CFA) 1.0, has been widely used in many commercial cameras. As shown in Figure 1a, for each 2 × 2 block, there are two green pixels, one red pixel, and one blue pixel. Even in the Mastcam onboard the Mars rover Curiosity [2,3,4,5], the Bayer pattern has been used for the RGB images. The main reason for using the Bayer pattern is to reduce cost. Some researchers also developed image tamper detection algorithms based on demosaicing artifacts [6]. Because of the success of the Bayer pattern, a follow-up pattern, known as red-green-blue-white (RGBW) or CFA 2.0, was introduced by researchers at Kodak [7,8]. For each 4 × 4 block in a RGBW pattern (Figure 1b), there are 50% white pixels, 25% green pixels, and 12.5% red and blue pixels. In the past two decades, there are numerous other CFA patterns mentioned in [9,10,11,12,13,14,15].
For images collected in normal illumination conditions, our earlier papers [16] concluded that CFA 1.0 is better than CFA 2.0. In the CFA research community, one common belief is that CFA 2.0 has better performance for images taken in low lighting conditions. The argument is that, due to the presence of 50% white pixels in CFA 2.0, the signal-to-noise (SNR) of the whole image should be higher, hence the demosaicing performance of CFA 2.0 should be better for low lighting images.
From the above discussions, one may have several natural questions regarding the various CFA patterns. First, has anyone carried out a comparative study to compare CFA 1.0 and CFA 2.0 for low lighting images? To the best of our knowledge, only one paper [17] briefly mentioned that CFA 2.0 has some advantages in some images. This means that the claim that CFA 2.0 is more suitable for low lighting conditions may be based on intuition rather than on observations based on objective experiments. It will be good to have some objective measures to judge which CFA is better for low lighting conditions. Second, how can one perform objective experiments for low lighting conditions? If one collects images in low lighting conditions, then we may not have the ground truth images, which would be used to generate objective metrics. In [17], low lighting images were emulated by adding noise to clean images. It is well-known that the noise induced by low lighting conditions is called Poisson noise, which is magnitude-dependent. If one simply adds Gaussian noise to clean reference images, then the noise behavior will be very different from that of images collected in actual low lighting conditions. In [18], a procedure for adding Poisson noise is mentioned in detail, and we have adopted that procedure in this research. Third, after the demosaicing of low lighting images, the demosaiced images are still noisy. A common practice is to perform some denoising and contrast enhancement to improve the image quality. One immediate question regards where we should apply the denoising step. There are two places to introduce the denoising: after demosaicing and before demosaicing. Which one is better? Answering the above questions will have important implications in practice. First, practitioners or camera designers may design a camera in which a denoising algorithm is activated if lighting conditions are unfavorable. Second, camera manufacturers need to know where denoising should be performed if CFA 2.0 is chosen.
In this paper, we attempt to answer the questions raised earlier. In Section 2, we will briefly review a number of demosaicing algorithms for CFA 1.0 and CFA 2.0. The algorithms range from conventional to deep learning. Two out of 16 methods for CFA 1.0 are deep learning algorithms. Although there are other promising learning methods in the literature [19,20,21,22], some serious customizations may be needed. In Section 3, we will summarize a comparative study that compares the performance of CFA 1.0 and CFA 2.0 using a benchmark data set (Kodak). Noisy images emulating two low lighting conditions were generated. The noisy images have 10 dBs and 20 dBs SNR. Three case studies were performed for each CFA: (1) no denoising; (2) denoising before demosaicing; (3) denoising after demosaicing. Our first major finding is that CFA 2.0 indeed helped the demosaicing performance for both 10 dBs and 20 dBs conditions. Our second finding is that the demosaicing performance of the CFAs performs even better if denoising is applied. Our third finding is that denoising before demosaicing is better than denoising after demosaicing. Our last finding is that denoising helps demosaicing more in the 10 dBs SNR case than the 20 dBs SNR case. Some discussions are included to provide some qualitative analysis for the above findings. Section 4 concludes the paper with some further remarks and future research directions.

2. Demosaicing Algorithms

In this section, we present some algorithms for demosaicing CFA 1.0 and CFA 2.0.

2.1. Algorithms for Demosaicing CFA 1.0

The following algorithms were evaluated in our experiments and they are briefly summarized below:
  • Linear Directional Interpolation and Nonlocal Adaptive Thresholding (LDI-NAT): This algorithm is simple but the non-local search is time consuming [23];
  • Malvar–He–Cutler (MHC): This is the algorithm in [24]. This is the default method for demosaicing Mastcam images [2] used by NASA. The algorithm is very efficient and simple to implement;
  • Directional Linear Minimum Mean Square-Error Estimation (DLMMSE): This is the Zhang and Wu algorithm in [25]. This method was investigated in Bell et al.’s paper [2];
  • Lu and Tan Interpolation (LT): This is a frequency domain approach [26];
  • Adaptive Frequency Domain (AFD): This is a frequency domain approach from Dubois [27]. The algorithm can also be used for other mosaicing patterns;
  • Alternate Projection (AP): This is the algorithm from Gunturk et al. [28];
  • Primary-Consistent Soft-Decision (PCSD): This is Wu and Zhang’s algorithm from [29];
  • Alpha Trimmed Mean Filtering (ATMF): This method is from [30,31]. At each pixel location, we demosaic pixels from seven methods; the largest and smallest pixels are removed and the mean of the remaining pixels are used;
  • Demosaicnet (Demonet): In [32], a feed-forward network architecture was proposed for demosaicing. There are D + 1 convolutional layers and each layer has W outputs and uses K × K size kernels. An initial model was trained using 1.3 million images from Imagenet and 1 million images from MirFlickr. Additionally, some challenging images were searched to further enhance the training model. Details can be found in [32];
  • Fusion using three best (F3) [30]: The mean of pixels from demosaiced images of the three best individual methods were used;
  • Bilinear: Bilinear interpolation is the simplest algorithm that uses the nearest neighbors for interpolation;
  • Sequential Energy Minimization (SEM) [33]: A deep learning approach based on sequential energy minimization was proposed in [33]. The performance was reasonable, except that the computation takes a long time due to sequential optimization;
  • Exploitation of Color Correlation (ECC) [34]: The authors of [34] proposed a scheme that exploits the correlation between different color channels much more effectively than some of the existing algorithm;
  • Minimized-Laplacian Residual Interpolation (MLRI) [35]: This is a residual interpolation (RI)-based algorithm based on a minimized-Laplacian version;
  • Adaptive Residual Interpolation (ARI) [36]: ARI adaptively combines RI and MLRI at each pixel, and adaptively selects a suitable iteration number for each pixel, instead of using a common iteration number for all of the pixels;
  • Directional Difference Regression (DDR) [37]: DDR obtains the regression models using directional color differences of the training images. Once models are learned, they will be used for demosaicing.
It should be noted that F3 and ATMF are both pixel-level fusion methods. Details can be found in [30].

2.2. Algorithms for Demosaicing CFA 2.0

The baseline approach refers to a simple upsampling of the reduced resolution color image shown in Figure 2. The standard approach for CFA 2.0 is shown in Figure 2, which illustrates how to combine the interpolated luminance image with the reduced resolution color image to generate a full resolution color image.
In the paper [16] written by us, we proposed a pansharpening approach to demosaicing CFA 2.0. This approach is illustrated in Figure 3. The missing pixels in the panchromatic band are interpolated. At the same time, the reduced resolution CFA is demosaiced. We then apply pansharpening to generate the full resolution color image. There are many pansharpening algorithms that can be used. Principal Component Analysis (PCA) [39], Smoothing Filter-based Intensity Modulation (SFIM) [40], Modulation Transfer Function Generalized Laplacian Pyramid (GLP) [41], MTF-GLP with High Pass Modulation (HPM) [42], Gram Schmidt (GS) [43], GS Adaptive (GSA) [44], Guided Filter PCA (GFPCA) [45], PRACS [46] and hybrid color mapping (HCM) [47,48,49,50,51] have been used in our experiments. The list is a representative, if not exhaustive, set of competitive pansharpening algorithms.
Recently, we further improved the pansharpening approach by integrating it with deep learning. This approach was summarized in a recent paper [52]. The key idea is to apply deep learning to improve two steps. The first is the demosaicing of the reduced resolution CFA (see Figure 4) via deep learning. The second is the improvement of the interpolation of the pan band via deep learning. The particular deep learning algorithm is Demonet mentioned earlier. We have seen good performance improvement.
In addition to the above mentioned algorithms for CFA 2.0, we also applied least-squares luma-chroma demultiplexing (LSLCD) over [53] in our experiments.
We also have two fusion based algorithms known as F3 and ATMF, which were used in our earlier studies [16,30,52]. F3 fuses the three best performing algorithms and ATMF fuses seven high-performing algorithms.
It should be noted that algorithms for CFA 2.0 are much fewer than those of CFA 1.0. There may be promising machine learning algorithms [19,20,21,22] that have the potential to be applied to demosaicing of CFA 2.0.

3. Comparative Studies

In this section, we will answer the questions raised in Section 1. One of them is whether CFA 2.0 is indeed better than CFA 1.0 for low lighting conditions. The second is how to emulate low lighting images. The third is where denoising should be introduced. In short, we will answer the following question: which one of the two CFAs is the best method for images collected in low lighting conditions?
Since there are many possible algorithms for each CFA, our strategy is to first perform a comparative study for all the algorithms for each CFA using the same data set. We then compare the best methods from all the CFA studies. That is, we compare the best against the best, to select the best CFA.

3.1. Low Lighting Images and Denoising

We downloaded a benchmark data set (Kodak) from a website (http://r0k.us/graphics/kodak/) and selected 12 images, which are shown in Figure 5. It should be noted that this dataset is well-known and has been used by many authors in the demosaicing community such as [23,25,26,27,28,29]. These clean images will be used as reference images for objective performance metrics generation. Moreover, they will be used to generate noisy images that emulate low lighting conditions.
Emulating images in low lighting conditions is non-trivial. This is because noise introduced in low lighting images is known as Poisson noise. Unlike Gaussian noise, Poisson noise is amplitude dependent. That is, the amount of noise applied depends on the magnitude range of the image. To create a consistent level of noise close to our SNR levels of 10 dBs and 20 dBs, we created a loop where each image was rescaled between 1 and some number less than 255. Poisson noise was applied to each band. The rescaling was adjusted until the PSNR between the ground truth and the noisy image was as close to the desired SNR level as possible. This technique is described in [18]. The noisy images at 10 dBs and 20 dBs are shown in Figure 6 and Figure 7, respectively.
It should be noted that simply adding Gaussian noise to the clean image cannot emulate low lighting images. For example, we added Gaussian noise to the clean images to create images at 10 dBs SNR. The noisy images are shown in Figure 8. It can be seen the image characteristics are totally different as compared to the Poisson noisy images shown in Figure 6.
We adopted the well-known denoising algorithm known as BM3D (Block Matching 3D) [54] in our denoising experiments.

3.2. Performance Metrics

In this paper, we have used five performance metrics to compare the different methods and CFA patterns.
• Peak Signal-to-Noise Ratio (PSNR) [55]
PNSR is related to Root Mean Squared Error (RMSE). The RMSE of two vectorized images S (ground truth) and S ^ (prediction) is defined as
RMSE ( S , S ^ ) = 1 Z j = 1 Z ( s j s ^ j ) 2
where Z is the number of pixels in each image. The ideal value of RMSE is 0 if the prediction is perfect. If the image pixels are expressed in doubles with values between 0 and 1, then
PSNR = 20 log ( 1 / RMSE ( S , S ^ ) )
A higher PSNR means better quality. A combined PSNR is the mean of the PSNRs of the R, G, B bands.
• Structural SIMilarity (SSIM)
This metric was defined in [56] to reflect the similarity between two images. The SSIM index is computed on various blocks of an image. The measure between two blocks x and y from two images can be defined as
S S I M ( x , y ) = ( 2 μ x μ y + c 1 ) ( 2 σ x y + c 2 ) ( μ x 2 + μ x 2 + c 1 ) ( σ x 2 + σ x 2 + c 2 )
where μ x and μ y are the means of blocks x and y, respectively; σ x 2 and σ y 2 are the variances of blocks x and y, respectively; σ x y is the covariance of blocks x and y; and c 1 c 2 are small values (0.01, for instance) to avoid instability. The ideal value of SSIM is 1 for perfect prediction.
• Human Visual System (HVS) metric
The HVS metric in dB is defined as
H V S = 20 l o g ( 255 / M S E H )
where
M S E H = K i = 1 I 7 j = 1 J 7 m = 1 8 n = 1 8 ( ( X [ m , n ] i j X [ m , n ] i j e ) T c [ m , n ] ) 2
I and J denote image size, K = 1 / [ ( I 7 ) ( J 7 ) 64 ] , X i j are the discrete cosine transform (DCT) [57] coefficients of 8 × 8 image block for which the coordinates of the its upper left corner are equal to i and j, X i j e are the DCT coefficients of the corresponding block in the original image, and T c is the matrix of correcting factors [58].
• HVSm (HVS with masking)
This metric is similar to HVS except that visual masking effects are taken into account. The inclusion of a block containing contrast masking is the only difference between HVS and HVSm. Details can be found in [59].
On the website of the authors of [59], there is a table containing the correlation of different metrics with human perception. For completeness, we include that table below (Table 1). It can be seen that HVSm and HVS have much higher correlation with human perception than PSNR and SSIM in terms of Spearman and Kendall correlation coefficients.
In addition to PSNR, SSIM, HVS, and HVSm, we also used CIELAB [65] for assessing demosaicing performance.
Before we summarize the detailed experimental results, we would like to use a diagram (Figure 9) to highlight the various studies and their corresponding sections.

3.3. CFA 1.0 Results

In this section, we summarize the CFA 1.0 studies for two SNRs: 10 dBs and 20 dBs. Within each SNR, we have three sub-cases. Both objective and subjective evaluations have been used in our studies.

3.3.1. 10 dBs SNR Case

We have three case studies. The first case is about demosaicing the noisy images without any denoising. The second case deals with the scenario where denoising is performed after demosaicing. The third case is to investigate the performance of denoising before demosaicing.
• Case 1: No Denoising
As mentioned earlier, we have 16 methods for demosaicing CFA 1.0, which were mentioned in Section 2.1. The F3 fusion method fuses the results of Demonet, Bilinear, and SEM, which were the best performing demosaicing methods. The ATMF fusion method uses the seven highest performing methods, which are Demonet, Bilinear, SEM, PCSD, DLMMSE, LDI, and LT. Table A1 in Appendix A summarizes the PSNR scores, which are the average of the three individual PSNR scores for R, G, and B bands, the CIELAB scores, SSIM, HVS, and HVSm metrics. One can see that all methods achieved PSNR values of around 16 dBs. All the SSIM values are low and the CIELAB scores are high (poor). The HVS and HVSm metrics are also not high.
Figure 10 shows the averaged PSNR, CIELAB, SSIM, HVS, and HVSm scores of all the 16 methods. There are some minor variations in the metrics.
Figure 11 shows the demosaiced results of Image 1 and Image 8. The demosaiced images have color distortion and noise.
In short, without denoising, all the demosaicing algorithms performed not so well at 10 dBs.
• Case 2: Denoising after Demosaicing
Here, our focus is to investigate the demosaicing performance with help from the BM3D denoising algorithm, which is applied after demosaicing is completed.
The F3 fusion method fuses the results of Demonet, Bilinear, and SEM, which were the best performing demosaicing methods in this case. The ATMF fusion method uses the seven highest performing methods, which are Demonet, Bilinear, SEM, DLMMSE, LDI, AP, and LT. Table A2 in Appendix A summarizes the PSNR, CIELAB, SSIM, HVS, and HVSm metrics. One can see that all methods achieved PSNR values of around 20 dBs, which are 4 dBs higher than those values in Table A1 in Appendix A. All the SSIM, CIELAB, HVS, and HVSm values have been improved over the no-denoising case.
Figure 12 shows the averaged PSNR, CIELAB, SSIM, HVS, and HVSm scores of all the 16 methods. All the scores have improved quite a lot, as compared to those in Figure 10.
Figure 13 shows the demosaiced results of Image 1 and Image 8. The demosaiced images look much better than those images in Figure 11. The artifacts are less noticeable after denoising.
• Case 3: Denoising before Demosaicing
Here, denoising was performed before demosaicing started. In other words, BM3D was applied to the CFA patterns before feeding them into the demosaicing algorithms. Intuitively, this makes more sense in practical applications because denoising should be more effective if one suppresses noise at the early stages rather than near the end of the demosaicing process.
The F3 fusion method fuses the results of Demonet, AP, and LT, which were the best performing demosaicing methods in this case. The ATMF fusion method uses the seven highest performing methods, which are Demonet, AP, LT, DLMMSE, DDR, LDI, and ECC. Table A3 in Appendix A summarizes the PSNR, CIELAB, SSIM, HVS, and HVSm metrics. One can see that all methods achieved metrics slightly better than those in Table A2 in Appendix A.
Figure 14 shows the averaged PSNR, CIELAB, SSIM, HVS, and HVSm scores of all the 16 methods. All the scores have improved slightly as compared to those in Figure 12.
Figure 15 shows the demosaiced results of Image 1 and Image 8. The demosaiced images look much better than those images in Figure 11. However, it is hard to visually judge whether images in Figure 15 are of a better quality than those in Figure 13.
From the above studies, one can easily make two observations. First, denoising plays a very important role in enhancing the overall demosaicing performance in low lighting conditions. In terms of PSNR, the improvement exceeds 10 dBs. Second, denoising before demosaicing starts is more effective than after demosaicing. We can observe one to two dBs of performance gain in PSNR.

3.3.2. SNR at 20 dBs

One may argue that the noisy low lighting images at 10 dBs may be too extreme because people seldom take pictures without flash lights in such dark conditions. Now, we investigate the performance of CFA 1.0 in more realistic low lighting conditions of 20 dBs. Similar to Section 3.3.1, we also have three sub-cases.
• Case 1: No Denoising
We have 16 methods for demosaicing CFA 1.0. The F3 fusion method fuses the results of Demonet, ARI, and LDI, which are the best performing demosaicing methods. The ATMF fusion method uses the seven highest performing methods, which are Demonet, ARI, LDI, Bilinear, LT, MLRI, and SEM. Table A4 in Appendix B summarizes the PSNR scores, which is the average of the three individual PSNR scores for R, G, and B bands, the CIELAB scores, SSIM, HVS, and HVSm metrics. It should be noted that some methods (SFIM and HPM) did not perform well. Other methods achieved PSNR values of around 22 dBs.
Figure 16 shows the averaged PSNR, CIELAB, SSIM, HVS, and HVSm scores of all the 16 methods. There are some big variations in the metrics.
Figure 17 shows the demosaiced results of Image 1 and Image 8. The demosaiced images do not look good because of color distortion, noise, and contrast.
In short, without denoising, all the demosaicing algorithms did not perform well at 20 dBs. That is, the demosaiced images have the same quality as the input CFAs.
• Case 2: Denoising after Demosaicing
Here, our focus is to investigate the demosaicing performance with help from the BM3D denoising algorithm, which is applied after demosaicing is completed.
The F3 fusion method fuses the results of Demonet, bilinear, and ARI, which were the best performing demosaicing methods in this case. The ATMF fusion method uses the seven highest performing methods, which are Demonet, bilinear, ARI, LDI, AP, LT, and MLRI. Table A5 in Appendix B summarizes the PSNR, CIELAB, SSIM, HVS, and HVSm metrics. One can see that all methods achieved PSNR values of around 22 dBs, which are 2 dBs higher than those values in the Table A4 in Appendix B. All the SSIM, CIELAB, HVS, and HVSm values have all been slightly improved over the no denoising case.
Figure 18 shows the averaged PSNR, CIELAB, SSIM, HVS, and HVSm scores of all the 16 methods. All the scores have improved slightly as compared to those in Figure 16.
Figure 19 shows the demosaiced results of Image 1 and Image 8. The demosaiced images look slightly better than the images in Figure 17. The artifacts are less noticeable after denoising.
• Case 3: Denoising before Demosaicing
Here, denoising was performed before demosaicing started. That is, BM3D was applied to the CFA patterns before feeding them into the demosaicing algorithms.
The F3 fusion method fuses the results of Demonet, DLMMSE, and AP, which were the best performing demosaicing methods in this case. The ATMF fusion method uses the seven highest performing methods, which are Demonet, DLMMSE, AP, LT, ARI, LDI, MLRI, and ECC. Table A6 in Appendix B summarizes the PSNR, CIELAB, SSIM, HVS, and HVSm metrics. One can see that all methods achieved metrics slightly better than those in Table A5 in Appendix B.
Figure 20 shows the averaged PSNR, CIELAB, SSIM, HVS, and HVSm scores of all the 16 methods. All the scores have improved slightly as compared to those in Figure 18.
Figure 21 shows the demosaiced results of Image 1 and Image 8. The demosaiced images look much better than those images in Figure 17. However, it is hard to visually judge whether the images in Figure 15 are of better quality than those in Figure 19.
From the above studies, one can easily obtain two observations. First, denoising plays an important role in enhancing the overall demosaicing performance in low lighting conditions. In terms of PSNR, the improvement exceeds 2 dBs. Second, denoising before demosaicing starts is more effective than that of after demosaicing. We can observe one to two dBs of additional performance gain in PSNR if denoising is done before demosaicing.

3.4. CFA 2.0 Results

The objective of this section is to investigate the performance of CFA 2.0 in low lighting conditions. We have two SNR cases: 10 dBs and 20 dBs. Within each SNR case, we have three sub-cases, depending on whether denoising is applied or not.

3.4.1. SNR at 10 dBs

Here, we have three cases. The first case is about demosaicing the noisy images without any denoising. The second case deals with the scenario where the denoising is performed after demosaicing. The third case is to investigate the performance of denoising before demosaicing.
• Case 1: No Denoising
We have compared 15 methods for demosaicing CFA 2.0 pattern. Those methods are summarized in Section 2.2. The baseline refers to the bicubic interpolation of the reduced resolution color image shown in Figure 2. The F3 fusion method uses the three best performing methods, which are the Baseline, Standard, and GFPCA. ATMF uses the 7 best performing methods: Baseline, Standard, GFPCA, GSA, PCA, GS, and PRACS. From Table A7 in Appendix C, it can be seen that the averaged PSNR score of F3 yielded the best score, which is 21 dBs.
Figure 22 shows the average performance metrics of PSNR, CIELAB, SSIM, HVS, and HVSm. All of them are reasonable. Figure 23 shows the demosaiced images of three methods. One can still see some noticeable artifacts.
• Case 2: Denoising after Demosaicing
Here, denoising was applied after demosaicing. The F3 fusion method uses the three best performing methods, which were the Demonet+GFPCA, GFPCA, and LSLCD. ATMF uses the seven best performing methods: Demonet+GFPCA, GFPCA, LSLCD, Standard, PCA, GS, and PRACS. From Table A8 in Appendix C, it can be seen that the averaged PSNR score of LSLCD yielded the best score, which is more than 24 dBs. This is better than those numbers in Table A7 in Appendix C. The other metrics in Table A8 of Appendix C are all improved as well.
Figure 24 shows the average performance metrics of PSNR, CIELAB, SSIM, HVS, and HVSm. All of them look much better than those in Figure 22.
Figure 25 shows the demosaiced images of three methods. It can be seen that the artifacts in Figure 23 have been reduced quite a lot. Visually speaking, F3 has minimal distortion for the fence area of Image 8.
• Case 3: Denoising before Demosaicing
Here, denoising was applied before demosaicing. That is, the BM3D algorithm was applied to the CFA patterns. Intuitively, denoising before demosaicing should perform better that that of after demosaicing. The F3 fusion method uses the three best performing methods, which were the Standard, Demonet + GFPCA, GSA. ATMF uses the seven best performing methods: Standard, Demonet + GFPCA, GSA, HCM, GLP, GS, and PRACS. From Table A9 in Appendix C, it can be seen that the averaged PSNR score of Demonet + GFPCA yielded the best score, which is more than 26 dBs. This is at least 2 dBs better than those numbers in Table A8 in Appendix C.
Figure 26 shows the average performance metrics of PSNR, CIELAB, SSIM, HVS, and HVSm. All of them look much better than those in Figure 22 and slightly better than those in Figure 24.
Figure 27 shows the demosaiced images of three methods, not necessarily the best performing methods. It is difficult to judge whether or not the demosaiced images in Figure 27 is better than that of Figure 25.

3.4.2. SNR at 20 dBs

There are three case studies here.
• Case 1: No Denoising
There are 15 methods. The F3 fusion method uses the three best performing methods, which were the Baseline, Standard, and GFPCA. ATMF uses the seven best performing methods: Baseline, Standard, GFPCA, GSA, GS, PRACS, and LSLCD. From Table A10 in Appendix D, it can be seen that the averaged PSNR score of F3 yielded the best score, which is slightly above 20 dBs. The other metrics are mediocre.
Figure 28 shows the average performance metrics of PSNR, CIELAB, SSIM, HVS, and HVSm. All of them can be considered reasonable as demosaicing methods do not have denoising capability in general. Figure 29 shows the demosaiced images of three methods. One can see some artifacts.
• Case 2: Denoising after Demosaicing
Here, denoising was applied after demosaicing. The F3 fusion method uses the three best performing methods, which were the Demonet + GFPCA, GFPCA, and PRACS. ATMF uses the seven best performing methods: Demonet + GFPCA, GFPCA, PRACS, GSA, PCA, GS, and LSLCD. From Table A11 in Appendix D, it can be seen that the averaged PSNR score of LSLCD yielded the best score, which is 24.391 dBs. This is better than most of the PSNR numbers in Table A10 in Appendix D, but only slightly better than the LSLCD method (24.05 dBs) in Table A8 of Appendix C (10 dBs SNR case). This means denoising has more impact for low SNR cases than high SNR cases.
Figure 30 shows the average performance metrics of PSNR, CIELAB, SSIM, HVS, and HVSm. All of them look better than those in Figure 28.
Figure 31 shows the demosaiced images of three methods. It can be seen that the artifacts in Figure 29 have been reduced. However, some artifacts are still very noticeable, especially the color distortions. This means there is still room for further improvement in the future.
• Case 3: Denoising before Demosaicing
Here, denoising was applied before demosaicing. That is, the BM3D algorithm was applied to the CFA patterns. The F3 fusion method uses the three best performing methods, which were the Standard, GSA, and HCM. ATMF uses the seven best performing methods: Standard, GSA, HCM, GLP, GS, and HPM. From Table A12 in Appendix D, it can be seen that the averaged PSNR score of GSA yielded the best score, which is 28.172 dBs. This is 4 dBs better than those numbers in Table A10 in Appendix D, and 2 dBs better than the best method in Table A11 in Appendix D.
Figure 32 shows the average performance metrics of PSNR, CIELAB, SSIM, HVS, and HVSm. All of them look better than those in Figure 28 and slightly better than those in Figure 30.
Figure 33 shows the demosaiced images of three methods, not necessarily the best performing methods. It is difficult to judge whether or not the demosaiced images in Figure 33 are better than those of Figure 31 because some color distortions are still present.

3.5. Best Against the Best Comparison Among the Two CFA Patterns

Now, we would like to compare the two CFA patterns. Since different algorithms were used in each CFA, we think that an appropriate way to compare the different CFAs is to compare the best against the best. That is, for each CFA, we select the best performing method and compare its results with the best performing methods in the other CFA.
We have two case studies below: 10 dB SNR and 20 dB SNR. For each SNR, we have three sub-cases: no denoising, denoising after demosaicing, and denoising before demosaicing.

3.5.1. 10 dBs SNR

Table 2 and Figure 34 summarize the average performance metrics for the 10 dBs SNR case in our earlier studies in Section 3.2 and Section 3.3 In Table 2, the name of the best performing algorithm is also included in each cell alongside the metrics. From Table 2 and Figure 34, we have the following observations:
  • In the no denoising case, CFA 2.0 is indeed better than CFA 1.0. For instance, the PSNR gain in Figure 34a is more than 4 dBs, which is significant;
  • Denoising definitely improves the demosaicing performance, regardless of where the denoising is done. For CFA 1.0, the improvement over no denoising is about 4 dBs; for CFA 2.0, the improvement is more than 3 dBs in terms of PSNR. For other metrics in Figure 34b–e, we also observe big improvements;
  • Denoising before demosaicing has a better performance than that of denoising after demosaicing. For CFA 1.0, the improvement is 1.1 dBs and, for CFA 2.0, the improvement is 2.1 dBs in PSNR.

3.5.2. 20 dBs SNR

Table 3 and Figure 35 summarize the best against the best results for different CFAs under different denoising/demosaicing scenarios. We have the following observations:
  • In the no-denoising case, CFA 2.0 is 2.8 dBs better than CFA 1.0 in terms of PSNR (Figure 35a). Other metrics in Figure 35b–e also improved quite significantly;
  • Denoising definitely helps the demosaicing performance, regardless of where the denoising is done. For CFA 1.0, the improvement is over 2 to 3.5 dBs; for CFA 2.0, the improvement is more than 1.1 to 4.8 dBs in terms of PSNR. There are also big improvements in other metrics (Figure 35b–e);
  • Denoising before demosaicing has a better performance than that of denoising after demosaicing. For CFA 1.0, the improvement is 1.2 dBs and, for CFA 2.0, the improvement is close to 4 dBs in PSNR;
  • Denoising helps the demosaicing performance more when the SNR is low. More than 4 dBs of gain in PSNR were observed after denoising in the 10 dBs SNR case;

3.5.3. Discussions

Here, we provide some qualitative analyses/explanations for some of those important findings in Section 3.5.1 and Section 3.5.2:
  • Why denoising before demosaicing is better that that of after demosaicing:
    One intuitive explanation is that noise can be suppressed more effectively earlier rather than later. Once noise has propagated to subsequent steps in the processing pipeline, it is harder to suppress it because some steps in the demosaicing process may be nonlinear. For example, in deep learning approaches, some rectified linear units (ReLu) are inherently nonlinear. This intuition has been found to be valid in our past research on active noise suppression in noisy conditions, as well. For a NASA project on noise suppression in Space Station [66,67], we noticed that noise was suppressed more effectively near the source than farther away from the source, as there are more reflections in the far-field due to multipath propagations;
  • Why CFA 2.0 is better than CFA 1.0 in low lighting conditions:
    We believe a concrete theory is needed to explain why CFA 2.0 has better performance than CFA 1.0 and this could be a good future research topic. The inventors of CFA 2.0 also did not provide a theory behind this. Intuitively, we agree with the inventors of CFA 2.0 that this must have something to do with the amount of white pixels in CFA 2.0. According to the inventors of CFA 2.0, more white pixels improve the sensitivity of the imager. We offer another analysis below.
    We use the bird image at 10 dBs condition (Image 1 in Figure 6 of our paper) as a case study. There is no denoising in the demosaicing process. Figure 36 below contains two histograms of the residual images (residual = reference − demosaiced) for CFAs 1.0 and 2.0. From this figure, it can be seen that the histogram of CFA 2.0 is centered near zero, whereas the histogram of CFA 1.0 is biased towards the right, meaning that CFA 2.0 is more accurate (close to the ground truth), because of its better light sensitivity, than CFA 1.0;
  • Why denoising helps slightly more for 10 dBs case than the 20 dBs case:
    From Table 2 and 3, we noticed that the gap between denoising improvement in 10 dBs and 20 dBs is slim. However, we still noticed that denoising helps the demosaicing performance slightly more in the 10 dBs case than in the 20 dBs case. We do not have a concrete theory behind this. However, one intuitive explanation can be found using Figure 37, which is a hypothetical optimization problem. The x-axis shows the computational load and the y-axis shows the performance. This curve shows that, for the same amount of effort, the improvement in performance is higher in the early stage than the later. In other words, it is difficult to further improve once the system is already in good shape. In economics, there is a law of diminishing returns, which might be related to the case here.
    Although there is no physical law governing this behavior, we have seen similar observations in some engineering applications. For example, in a past paper on speech recognition [68] under noisy conditions, we noticed that the word recognition rate improves more when the SNR is low. See Table 1 in [68]. From that table, at 0 dB, the relative improvement is 140%, as compared to only 37% in the 6 dBs case. This implies that it may be easier to see improvements when a system starts from a poor condition.

4. Conclusions

In this research, we thoroughly investigated the performance of CFA 1.0 and CFA 2.0 for low lighting images. The low lighting images were emulated by introducing Poisson noise. We then applied more than 15 conventional and deep learning based algorithms to CFAs 1.0 and 2.0 using a set of emulated images at 10 dBs and 20 dBs SNR. Using both objective (five performance metrics) and subjective evaluations, we observed that the demosacing performance of CFA 2.0 is indeed better than that of CFA 1.0 in low lighting conditions. We also investigated where denoising should be performed. In our research, we experimented with two denoising scenarios: before and after demosaicing. One important observation is that denoising before demosaicing has a better performance than denoising after demosaicing. We also observed that denoising boosts the demosaicing performance more when the SNR is 10 dBs, compared to an SNR of 20 dBs.
In this paper, we have used the BM3D denoising algorithm, which is proven algorithm in the literature. In the future, other denoising algorithms may be tried. Moreover, we are exploring the possibility of incorporating CFA 2.0 in NASA’s future planetary missions to Mars and other planets. Lastly, we plan to investigate a direct approach to demosaicing CFA 2.0 using deep learning.

Author Contributions

C.K. conceived the overall concept and wrote the paper. J.L. implemented the algorithm, prepared all the figures and tables.

Funding

This research was supported by NASA JPL under contract # 80NSSC17C0035. The views, opinions and/or findings expressed are those of the author(s) and should not be interpreted as representing the official views or policies of NASA or the U.S. Government.

Conflicts of Interest

The authors declare no conflict of interest.

Appendix A. Performance Metrics of CFA 1.0 at 10 dBs. Three Cases: No Denoising, Denoising After Demosaicing, and Denoising Before Demosaicing

Table A1. Performance metrics of 16 algorithms for CFA 1.0 pattern at 10 dBs SNR. Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A1. Performance metrics of 16 algorithms for CFA 1.0 pattern at 10 dBs SNR. Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsDemonetPCSDDLMMSEDDRBilinearARILDIMHCAPAFDLTMLRIECCSEMF3ATMFBest Score
Img1PSNR17.21517.00117.05316.56116.40217.14517.09316.89416.22016.78117.09615.91517.16516.82516.99717.12217.215
Cielab11.32512.04611.51712.42014.08512.69111.72212.65212.70212.03511.68813.48512.01911.65611.39311.36911.325
SSIM0.2180.1950.1950.2030.2100.2500.2020.1940.1860.1880.2000.2040.2120.2990.3320.2350.332
HVS11.70311.57111.70711.08810.81311.60511.67311.43610.85711.41211.70210.39111.65111.28911.39711.61411.707
HVSm11.80011.68311.81511.19410.90911.67211.78411.56210.94211.51311.81310.47811.76611.33611.44611.69711.815
Img2PSNR15.49815.43515.46015.47515.33516.14815.45615.31615.39915.40615.46715.01615.48416.62415.97715.63816.624
Cielab10.89511.71611.22611.63013.35811.37811.50612.32111.33911.39911.45712.31211.7559.43810.48710.9979.438
SSIM0.4780.4680.4730.4750.3630.4330.4700.4610.4630.4690.4710.4650.4700.5210.5490.5060.549
HVS10.91210.78210.84810.82910.66111.50210.81510.66110.80710.77410.83610.34310.80212.00911.30110.96612.009
HVSm11.00710.88910.95210.94210.80511.61910.92310.77810.91110.87710.94310.44310.91112.11211.38611.05912.112
Img3PSNR17.39816.84516.88515.78318.92716.21116.28316.44217.42916.00916.05916.18015.20817.63718.26517.06218.927
Cielab11.45612.40411.79713.82512.29614.24012.95813.43111.12213.19913.26413.36014.97910.69810.23311.62510.233
SSIM0.3540.3300.3310.3290.3450.3540.3310.3300.3250.3180.3290.3310.3240.4240.4530.3730.453
HVS12.32111.75611.91510.66813.74011.04311.19811.34812.57210.95810.98111.06610.03512.52713.09011.92113.740
HVSm12.44911.89212.04910.77513.99411.12611.31211.48412.73311.06211.08711.18610.12112.61613.18712.02713.994
Img4PSNR14.87614.62514.85114.75514.57911.84314.86413.71214.72514.75214.87714.84114.49214.78915.20914.99215.209
Cielab13.13115.02313.77415.08418.71420.04114.68617.66414.11314.31514.47215.31815.79612.20512.26313.15812.205
SSIM0.4800.4740.4800.4780.3980.3700.4800.4520.4670.4710.4810.4770.4740.5010.5750.5200.575
HVS10.58310.14010.47110.2579.9297.16810.3849.13110.38110.37210.42510.3369.91610.34710.60610.45610.606
HVSm10.87110.43710.77510.56910.3007.32310.6969.38710.68410.67510.73510.66110.20110.59810.84010.72210.871
Img5PSNR17.38217.20417.23917.21217.28715.41517.26817.21517.12717.14517.27417.25717.32616.62017.33117.33917.382
Cielab8.9399.5619.1559.62111.23112.5219.46710.1719.2849.3649.4049.7629.8809.5849.1069.1128.939
SSIM0.2690.2610.2630.2650.2370.2590.2650.2580.2550.2570.2650.2630.2660.3110.3540.2930.354
HVS13.35813.18413.25113.12913.02711.24813.20113.10913.16713.15713.23313.14413.17212.49713.15113.24713.358
HVSm13.49613.35113.41113.30513.20811.32113.36513.30013.32513.31813.39513.32113.33412.57013.23213.37213.496
Img6PSNR18.29217.98618.08018.09718.34219.73718.11117.63617.76217.98318.12717.48417.55818.61218.81118.38219.737
Cielab11.49012.41411.41512.27814.73412.35911.93413.70511.86511.77811.88613.15013.1619.87310.24110.9859.873
SSIM0.3690.3500.3540.3580.3020.3460.3570.3450.3400.3460.3570.3500.3540.4210.4710.3970.471
HVS14.07513.83214.03213.90313.98515.59813.94313.39613.74413.95214.00513.24413.26214.30814.47714.13815.598
HVSm14.30814.10314.30014.19614.31715.92314.21413.66813.99414.21914.27613.48213.48914.49714.66114.36215.923
Img7PSNR17.90917.27417.69216.76517.98618.20617.72116.78917.39717.18017.73116.50917.49820.12718.98917.99820.127
Cielab10.01911.19010.15411.90012.89111.50410.64212.55310.55610.92210.55112.47911.4627.0598.7579.8437.059
SSIM0.3410.3220.3260.3240.2630.3120.3270.3140.3120.3190.3270.3200.3280.4100.4440.3670.444
HVS13.75613.15613.66212.56613.65814.01113.59412.57013.39213.10113.63612.28213.25916.04014.73713.80916.040
HVSm13.90313.31713.83812.71313.86214.14213.77212.73013.55613.25413.81512.41513.42316.20214.84513.94716.202
Img8PSNR16.82817.11717.17516.84417.03517.15516.88516.82515.95216.70016.87916.56316.33717.18317.25517.13517.255
Cielab10.68510.78810.20411.11212.81211.80210.91911.79311.80010.92510.87611.59711.9879.77010.07710.2699.770
SSIM0.3980.3870.3910.3880.3230.3680.3890.3800.3730.3830.3890.3830.3840.4320.4800.4250.480
HVS11.91112.20512.36111.96012.05512.22211.99711.92111.09411.84912.01411.64011.37612.25012.25712.17712.361
HVSm12.04812.38212.53412.13312.27812.37512.16312.10811.22412.00512.17811.80311.52312.36712.37412.31812.534
Img9PSNR12.72312.66712.68212.68013.55410.62312.68912.67513.20812.63312.69112.34612.70613.96813.48812.91513.968
Cielab11.75412.11711.81912.06512.19115.96811.98612.47711.11411.97011.95412.64612.2209.85910.68211.4689.859
SSIM0.2770.2700.2720.2710.2360.2390.2730.2680.2690.2690.2730.2690.2730.3030.3310.2950.331
HVS8.2598.1758.2238.1999.0426.1168.2118.1918.7748.1648.2227.8518.2059.5368.9988.4149.536
HVSm8.2988.2248.2698.2509.1146.1408.2588.2458.8288.2118.2697.8998.2539.5809.0318.4549.580
Img10PSNR16.97016.78116.54616.65116.31717.82216.85316.24315.82016.65616.71216.69116.88817.49217.14016.97417.822
Cielab10.16210.88910.70511.05213.34910.80810.70812.23411.68510.71910.81111.16411.0899.1259.92010.1919.125
SSIM0.3850.3760.3770.3780.3080.3690.3790.3680.3630.3730.3780.3760.3790.4160.4480.4050.448
HVS13.16012.98512.74412.78312.35414.00313.00212.35811.98712.87712.88112.81912.99113.58313.21113.10314.003
HVSm13.32413.18112.91812.97312.57314.21213.19412.54412.13613.06113.06613.01113.18213.73313.34613.26414.212
Img11PSNR16.63615.80415.49215.16016.10216.81916.24616.21815.10416.14216.12715.27716.56016.33916.52816.22316.819
Cielab11.65013.11013.15414.03914.26212.52812.28413.02113.84612.27212.41613.99012.27411.51711.51312.04411.513
SSIM0.3840.3620.3610.3600.3210.3580.3710.3640.3470.3590.3700.3600.3740.4190.4780.4110.478
HVS11.51210.66610.40710.00710.91711.70011.16011.11810.01611.09611.05010.13311.43511.18911.33211.06511.700
HVSm11.61310.76210.49210.09111.03411.79811.26711.24110.09211.19911.15210.22111.55511.26211.40011.14811.798
Img12PSNR16.76616.01316.68616.22816.62616.89316.68815.48815.54515.89016.69915.93116.06216.45716.83116.55416.893
Cielab10.96212.34811.00012.16613.33411.91911.37613.72312.64212.22911.30612.64312.52811.00510.71711.17210.717
SSIM0.4510.4290.4390.4310.3770.4270.4410.4200.4140.4220.4420.4290.4340.5020.5430.4790.543
HVS12.21011.38312.16711.61512.07012.36012.11110.86310.96511.32112.12911.31611.43911.85112.20811.92112.360
HVSm12.34611.51612.31911.75812.28412.51912.26710.99111.08211.44612.28311.45111.57411.95912.32312.04512.519
AveragePSNR16.54116.22916.32016.01816.54116.16816.34715.95415.97416.10616.31215.83416.10716.88916.90216.52816.902
Cielab11.03911.96711.32712.26613.60513.14711.68212.97911.83911.76111.67412.65912.42910.14910.44911.01910.149
SSIM0.3670.3520.3550.3550.3070.3400.3570.3460.3430.3480.3570.3520.3560.4130.4550.3920.455
HVS11.98011.65311.81611.41711.85411.54811.77411.34211.48011.58611.75911.21411.46212.28512.23011.90312.285
HVSm12.12211.81111.97311.57512.05711.68111.93511.50311.62611.73711.91811.36411.61112.40312.33912.03412.403
Table A2. Performance metrics of 16 algorithms for CFA 1.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A2. Performance metrics of 16 algorithms for CFA 1.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsDemonetPCSDDLMMSEDDRBilinearARILDIMHCAPAFDLTMLRIECCSEMF3ATMFBest Score
Img1PSNR18.63318.81719.05419.12120.65119.12718.54318.13422.08119.19518.73318.29718.48820.28020.22119.78522.081
Cielab9.1109.3718.8528.8367.8479.1999.41810.1086.6878.8839.2339.7089.5357.9347.6047.9036.687
SSIM0.3950.3660.3840.3660.3370.3490.3620.3390.3960.3770.3670.3550.3520.4090.4300.4330.433
HVS12.92613.11813.42313.43614.95613.50512.89312.46416.47813.52213.09612.62712.81414.72314.59714.14316.478
HVSm12.97913.18013.48413.50615.09413.58012.95312.52716.59513.58613.15712.68812.87814.79714.67214.20816.595
Img2PSNR18.08418.57818.96418.82619.94920.20518.46316.87918.99118.30118.57416.70018.00920.68519.56919.15420.685
Cielab7.8438.0027.3137.7647.4017.1217.8949.7787.3968.0037.7879.7128.4285.7826.5666.8395.782
SSIM0.5130.5130.5260.5150.3760.4080.5010.4850.5200.5200.5060.4840.4830.4640.4820.5310.531
HVS13.42413.84314.27614.12715.36715.67513.77012.16114.28413.54613.89211.98113.30816.17715.01914.48616.177
HVSm13.55014.00014.43814.29315.73215.97513.92912.28114.45213.68814.05012.09013.45416.43515.22314.64716.435
Img3PSNR21.40121.13220.61519.75622.39519.26019.81420.23723.35120.16719.44820.10918.11122.52122.14921.48023.351
Cielab7.1377.7087.6938.5757.1939.6398.4468.4796.2428.2478.7678.31310.1866.4036.4416.7216.242
SSIM0.5220.5010.5110.4880.4540.4500.4880.4760.5210.5000.4880.4810.4590.5230.5390.5510.551
HVS16.06915.69015.34914.45016.98714.01914.51914.89618.00914.83714.17814.79712.83017.41216.95216.21718.009
HVSm16.22115.84915.48814.57917.30414.15414.64815.05218.26414.96614.29614.93812.92717.62717.14516.37118.264
Img4PSNR17.28715.22215.44015.06616.57412.41116.64014.17916.88916.39116.78815.28514.71617.13017.37117.09417.371
Cielab10.37520.85820.57614.18614.89918.28912.42316.29011.64512.26012.12414.19021.1419.3669.5509.7359.366
SSIM0.5690.4870.4950.5250.4460.3800.5470.4930.5410.5410.5520.5230.4630.6010.6320.6260.632
HVS12.92411.45311.76710.39711.8497.73212.0319.49312.45711.86912.21210.62010.84012.57012.77212.48812.924
HVSm13.30311.93812.27510.66612.3817.90312.4069.73612.85512.22012.59510.91111.26912.88313.10612.79313.303
Img5PSNR23.49822.84222.67122.97121.83218.07423.19820.25919.93322.01823.29923.44623.00119.36021.71222.13723.498
Cielab4.3705.1294.9955.1406.0528.5514.9446.8346.7035.4704.8744.9795.1996.8435.2365.0204.370
SSIM0.3600.3510.3590.3450.2840.2850.3460.3300.3450.3530.3480.3410.3330.3260.3550.3750.375
HVS19.20918.51818.36918.55017.36513.87418.82815.90715.63417.65818.95718.98618.58015.20217.51417.91419.209
HVSm19.43018.75218.57218.80017.67813.98119.08216.06815.75517.84419.21019.27118.84515.30317.68718.08119.430
Img6PSNR22.62621.70522.78521.61723.18522.76421.43820.44219.33723.25721.92220.44021.62422.59223.19322.54323.257
Cielab6.8428.0936.7958.0707.8528.1057.8939.2439.2936.9717.5908.8578.0996.3596.0556.3696.055
SSIM0.4280.4380.4510.4350.3470.3490.4280.4170.4020.4470.4330.4160.4100.3590.4170.4500.451
HVS18.16817.18518.36517.12018.62618.54916.97415.93314.90518.75317.49815.99717.16418.28819.00418.16319.004
HVSm18.56117.54218.80717.48619.37119.16017.32116.23815.12519.24417.88016.27717.53918.77719.53418.56819.534
Img7PSNR25.62125.07626.87024.03024.84923.87125.45321.86625.89426.06625.88124.38424.48226.86826.52826.79926.870
Cielab4.3285.2074.3075.7945.9766.1765.0887.1024.7814.8184.8985.7315.6703.4943.7673.7583.494
SSIM0.4500.4400.4590.4290.3120.3280.4270.4020.4430.4520.4330.4210.4070.4060.4360.4790.479
HVS21.23620.60622.44019.46420.14319.64820.94317.34121.34121.45421.41319.80319.91523.05922.54622.63123.059
HVSm21.67721.04723.07619.83120.76920.08221.44017.60021.85321.97421.95420.21420.34023.84223.19523.22223.842
Img8PSNR22.58121.26221.66821.48122.36221.28922.44119.94322.15121.20222.10921.08721.98023.57423.13622.46023.574
Cielab5.5006.8046.2026.6746.6977.2416.0238.0046.0976.7056.1626.9316.4394.7905.0675.3514.790
SSIM0.4800.4800.4910.4720.3800.3960.4760.4550.4870.4840.4770.4630.4550.4400.4780.5130.513
HVS17.47016.06216.58516.39917.30616.34517.35414.84417.05516.06017.05415.99416.88918.73318.24517.41218.733
HVSm17.74116.30416.83916.65917.88216.66717.67815.06017.34616.29417.35016.24017.19019.11518.62417.68719.115
Img9PSNR18.52216.33918.43917.02518.98211.08819.31615.18117.82217.54418.47516.82315.97318.08218.71218.47619.316
Cielab5.9077.7906.1007.2106.38814.5955.7219.0016.5806.8006.1727.3978.1066.1605.7585.8375.721
SSIM0.3220.3140.3260.3150.2620.2240.3200.3040.3200.3210.3200.3110.3000.2890.3110.3330.333
HVS14.04611.77513.92512.48914.4436.57714.78710.64613.28812.98513.96212.28411.44213.64314.26513.99014.787
HVSm14.10711.82613.99512.54814.6026.60114.87810.69313.35313.04614.03612.34311.49313.70114.34014.05314.878
Img10PSNR19.38819.72419.68318.73419.94220.91919.22717.06017.77019.75819.13517.30219.04720.64420.21819.62420.919
Cielab7.4057.6897.3848.4807.9537.2837.93810.4519.1527.5317.9809.8638.2446.2606.6097.0486.260
SSIM0.4290.4350.4390.4220.3400.3720.4230.4020.4170.4390.4250.4040.4110.3970.4220.4450.445
HVS15.43915.77815.72514.73815.92317.08515.23713.07813.76715.77515.17313.32515.07116.72316.34215.66517.085
HVSm15.64116.02915.95914.93916.33817.50315.46613.23313.93416.01615.39413.48115.29517.02116.62215.88717.503
Img11PSNR19.17019.84519.22519.18920.75021.17019.86819.24519.49918.75520.11618.60519.97718.46719.53419.49221.170
Cielab8.2908.0968.2688.5507.6787.4407.8678.7978.1588.8637.6659.1287.9618.7887.7147.7507.440
SSIM0.4080.4400.4370.4260.3600.3600.4340.4340.4340.4290.4370.4190.4160.3230.4010.4480.448
HVS13.98514.57214.05313.98115.59316.12614.68914.03314.31013.53314.95613.40514.79413.34814.42914.34216.126
HVSm14.11014.72314.18314.11615.85516.38014.84614.18214.44813.65115.12013.52614.96313.47414.57514.47416.380
Img12PSNR18.29817.71117.76817.57416.58320.01317.84916.30417.47017.85717.90317.93017.78217.14617.56917.75820.013
Cielab8.7979.5609.3029.76511.6027.7139.46811.5489.6699.3749.2939.3909.55010.3579.5979.1907.713
SSIM0.5200.5140.5240.5030.3970.4510.5110.4810.5100.5200.5130.5070.4920.4610.4980.5330.533
HVS13.70813.01113.14712.90712.08815.66313.22411.64112.80613.21113.26613.29113.16012.63313.06413.17815.663
HVSm13.85913.16113.29313.05212.27215.99713.38111.76112.94913.36013.42113.45113.31912.76213.20813.31615.997
AveragePSNR20.42619.85420.26519.61620.67119.18320.18818.31120.09920.04320.19919.20119.43220.61320.82620.56720.826
Cielab7.1598.6928.1498.2548.1289.2797.7609.6367.7007.8277.7128.6839.0476.8786.6646.7936.664
SSIM0.4500.4400.4500.4370.3580.3630.4390.4180.4450.4480.4420.4270.4150.4160.4500.4760.476
HVS15.71715.13415.61914.83815.88714.56715.43813.53615.36115.26715.47114.42614.73416.04316.22915.88616.229
HVSm15.93215.36315.86715.03916.27314.83215.66913.70315.57815.49115.70514.61914.95916.31116.49416.10916.494
Table A3. Performance metrics of 16 algorithms for CFA 1.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A3. Performance metrics of 16 algorithms for CFA 1.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsDemonetPCSDDLMMSEDDRBilinearARILDIMHCAPAFDLTMLRIECCSEMF3ATMFBest Score
Img1PSNR19.81319.77219.80719.77920.50219.80719.81319.83320.62419.75621.41019.81719.80921.54520.80019.96221.545
Cielab7.8977.9607.8397.8277.9977.8777.8247.8527.8037.89110.1377.8307.8467.3528.1237.7757.352
SSIM0.4020.3970.4060.4070.4010.4000.4060.4080.4210.4030.2110.4080.4020.4130.3610.4150.421
HVS14.09914.09714.15814.12616.14114.17114.16214.18116.27514.08614.27014.17014.16316.03714.94814.36716.275
HVSm14.16114.16314.22314.18916.27914.23914.22814.24416.37514.15114.36314.23514.23016.12615.02114.43216.375
Img2PSNR21.60321.53321.57621.56820.46921.48421.52621.47521.07821.48521.32321.54721.54421.76521.47221.55521.765
Cielab5.4135.5535.3875.4537.1745.5655.4635.6746.0795.4668.7205.4745.4805.2176.1905.4965.217
SSIM0.5420.5390.5440.5430.3870.5200.5360.5320.5160.5410.4530.5370.5380.5120.5310.5420.544
HVS16.91716.83816.91416.92315.26616.88716.88716.88015.72316.78916.04616.88916.88917.02216.36516.81417.022
HVSm17.16917.10017.17617.18915.59717.16617.15817.16315.93817.04616.30917.15317.15317.29616.59117.06817.296
Img3PSNR24.19724.09724.15124.17024.85424.10624.16124.21025.50424.06424.29824.15624.15222.62726.06024.65326.060
Cielab5.7075.8355.6635.6619.1365.9055.6745.6738.8225.7259.9035.7045.6946.4947.4555.7405.661
SSIM0.5390.5350.5400.5440.5000.5320.5410.5420.5340.5360.4650.5400.5390.5340.5450.5490.549
HVS19.06018.94219.11219.13015.02219.11419.10919.16415.23318.94015.13719.11619.12417.52816.44918.43619.164
HVSm19.30319.19619.37019.38815.19019.38019.36819.42315.34919.19015.28319.37519.38417.70816.59118.65219.423
Img4PSNR17.51917.47617.52917.51416.41117.86917.53417.37516.82417.41216.88817.50717.55516.15217.61417.56517.869
Cielab9.60811.39510.49311.51215.53711.89011.26512.86911.39610.97312.25011.70911.74310.7289.18410.6209.184
SSIM0.5420.5430.5470.5470.4430.5290.5460.5330.5260.5400.5280.5440.5460.5430.6110.5600.611
HVS13.36413.06513.22813.07311.15113.35713.11112.90411.80913.09612.43613.04813.05011.69412.82113.10713.364
HVSm13.81513.54313.69513.55711.59713.84413.59013.42512.15913.55712.86113.53813.53211.99313.15713.55013.844
Img5PSNR23.82523.75523.80323.77222.39323.76023.80123.78422.80023.66825.10123.79423.78621.21924.10323.86925.101
Cielab4.6704.7704.6884.7027.0664.7584.6714.8106.6804.7576.7734.7014.7116.0005.6834.7124.670
SSIM0.3700.3690.3760.3700.2980.3560.3720.3710.3510.3710.3210.3650.3650.3480.3710.3730.376
HVS19.91419.89719.93319.88217.20919.91419.92919.93617.53919.77318.71319.92519.92317.30818.85819.74519.936
HVSm20.15720.15020.18120.13217.46420.17720.18220.19617.70620.02018.94720.17720.17517.44619.05319.98020.196
Img6PSNR20.46520.42020.46020.46823.03320.40620.44820.46023.92820.40523.88620.44320.44322.02723.00621.03123.928
Cielab7.8257.9117.7017.78010.4657.8997.7337.8889.8087.7757.5377.7937.7876.5817.7387.5746.581
SSIM0.4040.4030.4070.4060.3330.3860.4040.4050.4110.4050.3580.3990.3990.3930.4140.4080.414
HVS16.21716.16116.22516.21017.05416.22916.20516.26217.59016.16217.58316.22616.22917.80817.28816.47017.808
HVSm16.46316.41516.48116.47317.51616.49616.46316.52217.94716.41717.99316.48016.48518.16117.60616.73718.161
Img7PSNR27.42827.29527.40127.32922.51727.30827.37127.27922.84627.23821.82027.37327.37525.76224.51527.73627.736
Cielab4.4394.5374.4064.45113.5194.5274.4014.45813.2364.49714.0964.4724.4654.80210.0645.3954.401
SSIM0.4710.4700.4750.4750.3600.4680.4690.4650.4490.4710.3140.4740.4730.4610.4540.4780.478
HVS23.21223.15323.30323.21315.94623.23023.30223.28816.17623.08316.63323.24323.23421.42918.49822.11423.303
HVSm23.85823.80823.97223.88116.13123.90423.98624.00816.31123.72216.82723.90923.89921.86618.71322.60724.008
Img8PSNR22.12522.05722.11522.09223.68122.05622.08922.00325.34022.02324.41222.09622.09321.10924.82222.60225.340
Cielab5.6935.7955.6425.6889.1455.8585.6815.9308.4645.7155.4905.7095.7106.3145.9305.5435.490
SSIM0.4920.4910.4960.4930.3990.4760.4920.4890.4860.4920.4320.4910.4900.4520.4960.4940.496
HVS17.19917.10917.22317.22418.26717.22917.21317.13519.51817.10518.02317.23317.23316.25718.38917.46319.518
HVSm17.44117.36317.48117.47818.93017.49317.47417.41619.96417.36118.37917.48817.48816.46018.70317.72619.964
Img9PSNR18.29318.25318.27718.27820.98118.25718.27518.27221.27018.19022.08818.27418.27120.97220.42518.72122.088
Cielab6.1766.3026.1686.1825.1936.3456.1786.3174.5946.2455.6746.2116.2304.7035.1665.8744.594
SSIM0.3270.3240.3290.3280.2810.3130.3270.3260.3220.3240.2950.3260.3250.3110.3320.3290.332
HVS13.73813.65413.72813.73116.70913.73313.73113.74116.97913.62815.26913.72613.72716.46515.27814.05416.979
HVSm13.79413.71513.78713.78816.93013.79513.79013.80217.10213.68715.36613.78513.78616.55815.35514.11517.102
Img10PSNR22.46122.35022.43322.38219.97822.32122.41022.38220.44322.32921.57722.40422.40219.91321.58522.23822.461
Cielab5.4915.6575.5015.5638.0255.7115.5385.7517.0715.5936.9995.5835.5867.1196.1505.5815.491
SSIM0.4560.4550.4600.4580.3460.4380.4560.4560.4290.4560.4050.4540.4530.4220.4520.4580.460
HVS18.86018.76318.84618.80415.15018.81618.82218.84815.55918.72515.71918.81718.83016.09216.72518.30218.860
HVSm19.23219.16419.24119.19215.47219.23419.22519.26415.77319.11515.97219.21319.22516.31316.97318.64619.264
Img11PSNR20.35920.31820.35620.34720.12920.31620.35420.35320.40820.29122.18020.35020.34719.75421.04020.52422.180
Cielab7.1067.1617.0467.0359.4747.1067.0317.1149.1277.1056.7247.0587.0597.6067.2706.9956.724
SSIM0.4140.4140.4180.4140.3300.3950.4150.4200.4000.4160.3400.4120.4100.3650.4070.4150.420
HVS15.19815.15415.23615.21712.77815.23315.23615.26012.90315.14615.89715.24015.23914.59014.68515.18015.897
HVSm15.35215.31215.39515.37612.90415.39915.39515.41913.00115.30316.11115.39915.39914.73514.82415.33616.111
Img12PSNR18.79518.73318.77418.82417.01618.69118.74718.75317.33418.72418.60518.75318.75117.79018.30018.69918.824
Cielab8.5128.6538.5038.50413.2688.6408.5248.61912.6798.58612.1338.5578.5549.66110.5798.8448.503
SSIM0.5370.5340.5400.5360.4010.5150.5360.5400.4910.5370.4450.5340.5340.5120.5200.5360.540
HVS14.05813.97914.07214.09610.51714.08214.05114.10910.65014.00013.26414.03914.05613.02912.60613.83714.109
HVSm14.21114.14314.23414.25810.64214.26014.21714.28010.73314.16113.42114.20414.22013.15912.72013.99014.280
AveragePSNR21.40721.33821.39021.37720.99721.36521.37721.34821.53321.29921.96621.37621.37720.88621.97821.59621.978
Cielab6.5456.7946.5866.6979.6666.8406.6656.9138.8136.6948.8706.7336.7396.8817.4616.6796.545
SSIM0.4580.4560.4610.4600.3730.4440.4580.4570.4450.4580.3810.4570.4560.4390.4580.4630.463
HVS16.82016.73416.83116.80215.10116.83316.81316.80915.49616.71115.74916.80616.80816.27216.07616.65716.833
HVSm17.08017.00617.10317.07515.38817.11617.09017.09715.69616.97815.98617.08017.08116.48516.27616.90317.116

Appendix B. Performance Metrics of CFA 1.0 at 20 dBs. Three Cases: No Denoising, Denoising After Demosaicing, and Denoising Before Demosaicing

Table A4. Performance metrics of 16 algorithms for CFA 1.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A4. Performance metrics of 16 algorithms for CFA 1.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsDemonetPCSDDLMMSEDDRBilinearARILDIMHCAPAFDLTMLRIECCSEMF3ATMFBest Score
Img1PSNR20.45520.31620.00219.99420.02920.71320.15620.00319.90819.93920.15520.05620.09220.07220.52020.39420.713
Cielab7.1867.8487.8267.9858.9758.0297.8528.4057.9467.9827.8288.0198.1198.6187.4317.4367.186
SSIM0.3210.3000.3000.3110.3150.3620.3080.2980.2890.2910.3070.3140.3190.3660.3470.3790.366
HVS15.05614.90914.63814.54114.48715.20614.73114.54914.59914.56514.75014.57314.58914.84615.03314.91015.206
HVSm15.17315.04614.75814.67114.63115.30514.85714.68914.71514.68314.87414.70814.71714.92915.13815.00115.305
Img2PSNR20.25220.14420.18920.19019.66420.39220.17220.06220.04420.08320.19120.18720.23420.20620.41120.44520.392
Cielab6.0156.8356.3916.8048.8937.1326.7117.4766.5256.5656.6456.8606.9056.3396.3426.2536.015
SSIM0.5780.5710.5770.5790.4460.5390.5730.5610.5660.5730.5750.5740.5760.6070.5900.6190.607
HVS15.95615.64015.75215.68315.09315.87215.66315.52415.61115.60115.70015.64315.65215.47715.89915.83815.956
HVSm16.16115.87715.97615.92815.44616.10815.90115.80215.83315.82015.93415.88815.89115.64716.10716.02316.161
Img3PSNR21.41520.59520.62021.11922.25920.45921.83320.18120.07320.64921.52621.38720.24720.23021.36321.32222.259
Cielab6.4727.7097.3557.2498.0718.5306.7198.3887.8057.4156.8757.1618.0587.7216.9866.8566.472
SSIM0.4550.4470.4470.4570.4450.4780.4560.4460.4360.4410.4540.4560.4570.5020.4760.5040.502
HVS16.59815.53615.67816.10517.11415.33616.88315.10815.10615.69516.57816.36115.12915.15816.32516.23017.114
HVSm16.79315.71915.85516.32317.45215.47917.12815.29015.26215.87316.80216.59615.29515.27116.50116.38317.452
Img4PSNR17.89717.90417.96117.95217.47018.77118.01517.74117.71817.77918.02517.95218.07717.80618.60018.86018.771
Cielab9.64912.13810.89412.27316.49412.82411.94514.01311.23511.49211.69512.61412.6959.49610.4549.8219.496
SSIM0.5110.5170.5210.5210.4410.5130.5220.5060.5060.5110.5230.5190.5230.5140.5470.5820.523
HVS14.51313.88714.16013.85212.89914.49613.96013.52314.01813.97514.02113.80713.81214.16714.55914.64214.513
HVSm15.20214.59714.86114.57813.66415.18614.66514.29714.70514.66314.72514.53714.51714.82015.21515.23915.202
Img5PSNR20.34320.08520.11420.08520.01720.48320.14320.06719.96219.98820.14920.13220.20120.12120.45120.44920.483
Cielab6.5056.7426.4266.7758.1697.2106.6717.2486.5286.6086.6216.8746.9756.2296.5706.3526.229
SSIM0.3330.3280.3300.3320.2910.3340.3320.3230.3210.3240.3320.3300.3330.3620.3530.3800.362
HVS16.41016.09516.16916.03615.76816.37716.10015.98516.03316.03416.13716.04916.07716.12616.37916.34116.410
HVSm16.57816.30616.36816.25616.02716.53816.30516.22416.23316.23716.34116.26816.28316.25716.54116.48616.578
Img6PSNR22.55121.02520.38920.49522.46421.11020.61020.32820.28220.31020.39421.17320.39520.33521.51421.18122.551
Cielab6.1797.9257.9508.2428.8978.4458.0309.0348.0928.1658.1647.8198.4557.8067.3037.3216.179
SSIM0.5610.5490.5510.5550.4590.5170.5520.5380.5370.5440.5520.5520.5500.5920.5740.6070.592
HVS18.53016.73116.11616.16418.22916.88116.30416.02716.04816.04716.09616.88816.05715.89917.24716.84418.530
HVSm18.84716.99116.33116.39818.82217.15316.53816.27216.26316.26016.31517.15816.27916.06717.49917.05818.847
Img7PSNR20.41520.31920.35520.30321.10420.38720.35820.29021.17020.26220.36820.35720.18422.20320.46520.73822.203
Cielab6.7407.4927.1107.5238.6658.0007.4228.0576.5237.2717.3527.6007.8075.4577.1826.8875.457
SSIM0.4710.4660.4700.4710.3740.4570.4670.4550.4590.4640.4690.4690.4720.5190.4900.5220.519
HVS16.38516.17416.25216.11916.82416.18116.18516.08417.16416.14616.21216.15915.94218.08016.29416.51318.080
HVSm16.52916.34316.41316.29617.09816.31716.35316.28217.37116.30616.37816.33616.10118.25916.43016.64118.259
Img8PSNR20.34720.10820.16220.13619.65220.62020.15020.05820.02620.05920.16520.14720.19120.10220.46020.42120.620
Cielab6.6627.4187.0197.4069.3447.7827.3068.0397.1447.2037.2447.4997.5506.9647.0236.8866.662
SSIM0.5040.4950.5000.4980.4090.4850.4990.4870.4900.4940.5000.4960.5000.5370.5200.5500.537
HVS15.62615.23015.36415.29514.71615.76315.30515.19615.22515.25015.34515.28215.29515.07415.60615.49215.763
HVSm15.82515.45615.58115.52715.04315.99815.53215.45515.43915.46615.56815.51915.52615.22615.81115.67615.998
Img9PSNR20.20420.02320.05220.04620.05220.53420.10519.98719.93319.88620.10120.08020.19420.03420.39520.45520.534
Cielab4.9685.6555.2665.6427.2486.2345.5336.2435.3425.4675.4795.7785.8734.8485.3345.0884.848
SSIM0.3170.3120.3130.3140.2730.3070.3150.3080.3080.3100.3150.3140.3160.3400.3290.3490.340
HVS16.18415.78015.92015.76315.56316.19015.86015.66715.84015.72515.88715.74915.80115.68216.13516.11016.190
HVSm16.37116.01216.14116.00615.85016.34716.08615.93916.05915.94316.10915.99516.02815.83116.30316.26016.371
Img10PSNR20.22020.06220.12420.07719.69420.30820.12620.04420.00820.01520.14020.11720.15920.06720.37820.38720.308
Cielab6.6447.3186.9017.2989.3117.7927.1887.9707.0067.0887.1207.3687.4586.7766.9046.7526.644
SSIM0.4730.4640.4670.4670.3820.4530.4670.4580.4580.4630.4680.4650.4670.4890.4840.5070.489
HVS16.54016.28416.36916.25315.78516.52516.30116.20116.25216.28116.34116.27516.28616.18116.56816.51816.540
HVSm16.76016.55416.62216.51816.17916.79116.56516.50316.50616.53516.60216.54416.54816.36316.80416.73116.791
Img11PSNR20.54320.18020.21820.21819.98820.43620.24120.16920.08320.11920.25120.22820.26320.22120.50820.49320.543
Cielab7.0287.6667.3307.6089.1387.9467.5148.1117.4447.4747.4667.6837.7437.3837.2717.1197.028
SSIM0.5200.5130.5150.5190.4430.4960.5200.5100.5020.5080.5200.5170.5200.5580.5400.5780.558
HVS15.56615.11015.22815.14914.90715.36515.18315.11115.14515.12315.21215.15715.16215.08115.41615.36115.566
HVSm15.72415.28215.39315.32815.12715.52715.35315.30715.30615.28515.38015.34015.34015.21015.56615.49415.724
Img12PSNR20.55720.43620.49520.54419.83020.54220.47220.39520.35520.39820.48820.47820.50120.39020.69420.71420.557
Cielab6.5117.2256.8317.1969.1117.5167.1287.8276.9637.0057.0677.3017.3386.6996.7696.6546.511
SSIM0.5770.5740.5770.5760.4910.5650.5790.5680.5650.5690.5800.5770.5800.6320.5980.6320.632
HVS16.20315.91816.06016.02215.49216.20415.96815.91015.94015.96915.97815.95815.97915.81916.23016.17016.204
HVSm16.41516.16916.29916.27615.87816.47916.21716.19616.17816.20716.22216.21616.23116.00816.46016.37616.479
AveragePSNR20.43320.10020.05720.09720.18520.39620.19819.94419.96419.95720.16320.19120.06220.14920.48020.48820.488
Cielab6.7137.6647.2757.6679.3608.1207.5018.4017.3807.4787.4637.7157.9157.0287.1316.9526.713
SSIM0.4680.4610.4640.4670.3970.4590.4660.4550.4530.4580.4660.4650.4680.5020.4870.5170.517
HVS16.13015.60815.64215.58215.57315.86615.70415.40715.58215.53415.68815.65815.48215.63215.97415.91416.130
HVSm16.36515.86315.88315.84215.93516.10215.95815.68815.82315.77315.93815.92515.73015.82416.19816.11416.365
Table A5. Performance metrics of 16 algorithms for CFA 1.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A5. Performance metrics of 16 algorithms for CFA 1.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsDemonetPCSDDLMMSEDDRBilinearARILDIMHCAPAFDLTMLRIECCSEMF3ATMFBest Score
Img1PSNR24.06222.76022.47722.81924.43424.62723.71321.14220.80022.41423.88324.28923.97820.76724.54123.98124.627
Cielab4.7445.8415.8455.6735.0425.0035.2156.9357.0635.9895.1294.9515.1188.0524.5914.8804.591
SSIM0.5150.4770.4910.4770.4390.4570.4790.4540.4790.4840.4830.4750.4690.4730.5150.5070.515
HVS18.43117.09916.86717.16418.86419.06618.08315.49715.20216.75318.26318.61318.31315.44818.97318.34919.066
HVSm18.54617.20416.96117.26819.11319.24318.21015.58015.27116.84618.39218.76318.45715.51319.12918.47419.243
Img2PSNR21.27519.42719.99519.57221.73021.47720.33919.32120.23719.63519.96219.44519.38920.66321.73520.68621.735
Cielab5.3807.0636.3966.9416.1845.9776.3157.4296.3106.7536.5447.0177.0636.0345.3035.8625.303
SSIM0.6180.5890.6040.5910.4370.4950.5870.5770.6040.6030.5880.5830.5690.5690.5560.5980.618
HVS16.74014.71615.31314.89517.24216.99815.67214.62915.51414.88215.29314.75914.71415.82617.26516.06417.265
HVSm16.92514.86715.47115.04817.75617.31115.85914.79515.68515.02915.46114.91014.86815.99317.56216.25217.756
Img3PSNR24.50122.77922.95522.68025.66720.89022.37721.40722.62622.42122.67723.14221.53221.78123.65023.06025.667
Cielab4.7356.0115.7195.9454.9547.6356.0966.9565.9876.1365.9145.7086.6946.5525.2875.5184.735
SSIM0.6210.5870.6000.5870.5410.5430.5840.5670.5940.5920.5880.5800.5690.5960.6060.6130.621
HVS19.31817.43217.71917.40020.39415.67617.09316.13917.34517.11617.40517.85716.26716.61118.45117.79520.394
HVSm19.51117.59517.87817.56120.90315.81417.24816.27117.49917.25917.56518.03916.40416.73018.66217.95720.903
Img4PSNR18.64918.09818.30918.13918.01419.21118.45717.66418.29718.09118.25918.03718.30018.20419.62419.20719.624
Cielab9.05411.69210.44411.85414.84212.19911.35813.53510.67711.06911.24212.19312.1429.1088.8569.3468.856
SSIM0.5470.5490.5560.5520.4620.5210.5560.5320.5390.5430.5560.5480.5520.5490.6140.6150.615
HVS15.00713.74114.14913.72513.32814.83214.10813.14614.30713.93613.91913.58613.76614.18415.21914.79315.219
HVSm15.66514.33014.74514.33114.10915.52314.73813.75714.94414.52314.50614.18314.37114.73915.85415.34415.854
Img5PSNR24.00321.15321.27121.25423.86323.59421.86620.26822.70520.61821.47321.81021.95221.04224.02122.74424.021
Cielab4.0835.7765.5725.7164.7594.7385.3246.5494.8396.0665.5355.4215.3755.6464.1324.6904.083
SSIM0.4020.3950.4010.3910.3300.3540.3920.3810.4010.3950.3920.3880.3810.3820.3940.4100.410
HVS19.89816.92817.04916.99719.46519.43017.60916.00618.41916.35817.23917.54417.69416.89119.88518.54719.898
HVSm20.09317.06017.17117.13319.89619.69017.76116.12918.59016.47217.37617.69917.85716.99520.14818.71320.148
Img6PSNR24.50622.16922.69521.33524.96921.82723.09619.87422.48422.20723.03522.00721.41919.52724.00723.12724.969
Cielab4.9486.7016.0247.1705.8637.2375.9568.6886.2536.4975.9706.7347.1328.4695.3215.7034.948
SSIM0.6010.5850.5970.5780.4670.4740.5850.5510.5890.5920.5870.5770.5530.4930.5500.5890.601
HVS20.21017.73218.29516.91920.78517.59018.70415.49418.07917.78818.65617.62617.04215.05519.85018.80720.785
HVSm20.58118.00218.58217.14821.74717.92219.03315.67718.35918.04918.97517.88317.28115.21320.32319.12621.747
Img7PSNR25.50826.68827.69327.00926.07520.64126.12027.01928.62728.85226.94226.73726.30928.94224.04126.45628.942
Cielab3.8094.0223.4634.0144.7717.1544.1544.3533.3533.3513.8664.1064.2252.7514.6613.6532.751
SSIM0.5990.5710.5890.5680.4190.4450.5580.5460.5860.5900.5660.5600.5480.5670.5370.5920.599
HVS21.37822.40123.47922.67721.76816.40321.83022.52724.26124.48222.67822.40121.99124.92819.89822.26724.928
HVSm21.65422.83723.99323.15722.46816.54722.22323.10324.90725.14723.14122.85722.40925.63720.16722.64325.637
Img8PSNR24.16721.38221.54422.50823.01423.55921.92721.29221.95821.62921.62022.44122.69321.16723.89622.79624.167
Cielab4.4056.2955.9855.6655.8965.5145.8786.6795.8186.0586.0365.7135.5826.2234.7215.1654.405
SSIM0.5950.5670.5780.5690.4550.4900.5630.5530.5780.5760.5640.5630.5540.5280.5580.5840.595
HVS19.27516.30516.52117.52018.09718.73516.92416.27116.87116.56216.63217.44117.70916.04219.07317.82919.275
HVSm19.56116.50816.72117.76818.70619.13717.15516.49917.09216.76816.84517.69317.97716.20119.48018.08419.561
Img9PSNR20.71119.77119.65820.45320.26421.54719.91918.83620.09719.75219.46619.48319.38719.75220.93820.20821.547
Cielab4.5915.4155.2945.0565.5774.9235.2566.1425.1475.3555.4705.5595.6375.1104.6454.9014.591
SSIM0.3530.3470.3510.3470.2830.2990.3440.3390.3500.3510.3440.3420.3340.3250.3330.3540.354
HVS16.30515.19315.12915.88515.71717.10315.38414.27315.53215.16114.93814.92114.83915.11116.50215.72117.103
HVSm16.38415.27715.20415.97915.90717.25015.47014.35615.61815.24015.01415.00414.92215.17516.62015.80317.250
Img10PSNR21.65520.66320.81720.55821.27122.25820.90719.73721.07720.90820.92420.28220.62520.40321.92021.29322.258
Cielab5.5736.6556.3196.7186.8155.9876.4037.6756.2116.3696.3646.9086.7006.5495.5685.8825.568
SSIM0.5050.5060.5120.5030.4020.4510.5020.4930.5090.5120.5040.4960.4910.4780.4880.5150.515
HVS17.79916.73416.87716.61217.35318.49416.95415.78117.07916.96116.99216.33316.68716.40518.12317.40318.494
HVSm18.02516.96717.09616.82817.85618.90717.19715.98817.31617.18717.22916.54516.91816.58218.47217.65018.907
Img11PSNR21.98520.96821.13420.88523.98321.73320.82820.44820.30420.60221.07921.20720.85019.66022.73421.47723.983
Cielab5.8556.7926.4966.8165.4006.4746.7797.3537.1636.9586.5926.6146.8667.7995.3876.1815.387
SSIM0.5470.5530.5580.5480.4590.4570.5460.5490.5450.5520.5490.5480.5280.4790.5240.5500.558
HVS16.86915.73715.95615.69319.10116.66515.63815.26415.11615.38015.90116.03215.68214.45017.73116.33219.101
HVSm17.03815.88616.10615.83619.58816.89115.78515.40515.24215.51416.05316.18915.83314.56117.97616.49419.588
Img12PSNR21.31320.35220.36520.46620.56721.30320.14919.83220.13820.24520.23920.19120.08020.16921.27120.67321.313
Cielab5.8906.8776.6776.8317.0796.3206.9537.5386.9276.8766.8657.0427.1266.8905.9186.3615.890
SSIM0.6650.6440.6550.6400.5210.5610.6350.6300.6480.6500.6390.6340.6200.6230.6210.6520.665
HVS16.82015.70615.76615.85416.33117.02615.53915.24015.50815.62515.61515.58915.50415.55416.92816.13117.026
HVSm17.02315.90815.95616.05216.75817.37715.73615.43515.69615.81215.81015.78515.70015.72617.22816.33817.377
AveragePSNR22.69521.35121.57621.47322.82121.88921.64220.57021.61321.44821.63021.58921.37621.00622.69822.14222.821
Cielab5.2566.5956.1866.5336.4326.5976.3077.4866.3126.4576.2946.4976.6386.5995.3665.6785.256
SSIM0.5480.5310.5410.5290.4340.4620.5270.5140.5350.5370.5300.5250.5140.5050.5250.5480.548
HVS18.17116.64416.92716.77918.20417.33516.96215.85616.93616.75016.96116.89216.68416.37618.15817.50318.204
HVSm18.41716.87017.15717.00918.73417.63417.20116.08317.18516.98717.19717.12916.91616.58918.46817.74018.734
Table A6. Performance metrics of 16 algorithms for CFA 1.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A6. Performance metrics of 16 algorithms for CFA 1.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsDemonetPCSDDLMMSEDDRBilinearARILDIMHCAPAFDLTMLRIECCSEMF3ATMFBest Score
Img1PSNR23.03122.96123.01419.77923.69323.00823.02123.04224.01322.92923.77123.03223.02323.38623.36923.05224.013
Cielab5.3185.4285.3097.8277.0405.3685.2945.3466.8475.3535.9985.2975.3196.1795.4345.2845.284
SSIM0.5010.4940.5040.4070.4800.4920.5050.5040.5040.5000.3810.5030.5000.4760.5290.5060.529
HVS17.37417.32417.41414.12614.97817.43917.42517.44515.07017.32117.02117.43617.43317.99716.65917.39917.997
HVSm17.46717.42517.51314.18915.06617.54517.52517.54315.12317.42017.12017.53517.53518.10116.73017.49418.101
Img2PSNR23.06122.96223.02321.56821.89622.87422.94022.84022.95722.91522.83922.98822.98020.92723.16123.02823.161
Cielab4.4844.6734.5015.4536.4994.7004.6034.9245.1574.5726.7774.5804.5915.6154.4844.5114.484
SSIM0.6240.6210.6260.5430.4530.6010.6170.6100.6020.6230.5580.6210.6210.5800.6390.6230.639
HVS18.44518.28318.38116.92316.75118.34318.32718.26917.45018.23815.57518.35218.35416.19718.20618.37618.445
HVSm18.70918.56418.65717.18917.18718.65018.62218.59117.69218.51015.75118.63418.63716.37818.44218.64918.709
Img3PSNR26.20526.07526.14324.17025.31326.03626.16026.18626.08226.02326.71926.15126.14624.83627.20326.19627.203
Cielab4.1614.3404.1735.6619.2424.4984.1744.2548.9264.2227.7404.2044.1994.8255.0624.1794.161
SSIM0.6170.6110.6170.5440.5640.6030.6170.6160.5990.6120.5420.6140.6150.6040.6370.6170.637
HVS21.06420.86521.08319.13016.34021.06221.06621.13816.55320.92917.34921.07321.07419.71019.51221.02421.138
HVSm21.32121.13521.35419.38816.51421.36221.34321.41716.64821.19417.48821.35021.35419.90719.66821.28821.417
Img4PSNR18.52218.49518.55817.51417.84919.15818.59118.32518.31518.38918.52418.54018.63518.45019.43318.97719.433
Cielab9.01211.18610.10511.51215.35011.74711.02312.89110.49610.64510.96611.57311.6348.9037.75610.0417.756
SSIM0.5380.5420.5460.5470.4560.5350.5460.5310.5300.5380.5400.5440.5470.5450.6230.5720.623
HVS14.88314.35414.60513.07313.31114.82714.42414.02414.55014.42114.56414.30114.29914.62815.28714.78515.287
HVSm15.54815.04315.28413.55714.09715.51015.11314.77615.23815.08715.28315.01014.98915.25215.83415.42915.834
Img5PSNR24.68124.60224.65623.77223.83724.58424.64524.60824.47224.50222.41824.64024.63025.06724.63524.66325.067
Cielab3.9774.0804.0014.7025.7274.0953.9934.1675.2924.0755.4014.0204.0313.8664.0653.9833.866
SSIM0.4150.4130.4210.3700.3360.3920.4160.4150.3970.4160.3440.4060.4060.3940.4280.4120.428
HVS20.67320.65920.68119.88216.34820.66020.67020.64916.61620.50217.28120.66820.67021.21919.23820.62821.219
HVSm20.89420.89420.90920.13216.53720.90920.90620.89716.71920.73217.41320.90120.90321.45519.38920.85221.455
Img6PSNR23.60623.51023.58620.46825.36523.40323.54223.50027.20123.50226.69023.55123.54023.00224.85523.62727.201
Cielab5.3425.5215.2857.7806.1485.6205.3425.7214.7875.3597.6555.3905.4035.6784.7855.2914.785
SSIM0.5950.5920.5980.4060.4670.5650.5920.5870.5970.5970.5570.5910.5900.5500.6170.5930.617
HVS19.32419.17519.27116.21019.47819.26619.23819.26320.49819.18620.56319.28419.27418.59319.80519.34120.563
HVSm19.62319.50019.59416.47320.16819.63219.56919.60920.92819.50721.04619.60619.59918.87220.12919.65921.046
Img7PSNR28.72528.56128.68727.32924.07628.51528.56528.36624.86428.51424.88628.64828.65228.82729.35828.70429.358
Cielab3.2633.4153.2574.45111.4993.4083.3303.53911.0943.3319.5143.3243.3193.1425.0563.2903.142
SSIM0.6010.5960.6040.4750.4620.5880.5960.5870.5790.6000.4070.6010.6010.5860.6260.6020.626
HVS24.58724.44124.60923.21316.58124.52124.50724.34316.87324.35620.26524.54024.54424.64121.78624.50424.641
HVSm25.15925.03025.20323.88116.76625.13725.12625.01916.97524.92420.59625.13625.14025.26422.04625.08125.264
Img8PSNR24.79224.69224.78022.09224.47924.63524.72524.51126.67224.65026.79324.74724.74623.79325.55824.81226.793
Cielab4.2004.3204.1675.6885.4904.4354.2234.5834.2024.2376.0704.2444.2464.6973.9274.1783.927
SSIM0.5960.5910.5980.4930.4710.5680.5930.5880.5830.5950.5450.5920.5920.5460.6140.5940.614
HVS20.00019.81219.96717.22418.15019.95219.94319.73519.38019.80018.57619.99020.00019.05219.90219.99920.000
HVSm20.31820.15320.31217.47818.74720.31620.29520.13519.71320.14518.87820.33320.34119.32520.20920.33320.341
Img9PSNR21.62421.55621.59518.27821.94921.54321.59021.56922.36021.45822.40421.58921.58322.13821.89621.61622.404
Cielab4.3234.4634.3326.1825.2514.5284.3424.5354.6064.4054.4684.3794.4004.0934.2674.3264.093
SSIM0.3600.3560.3610.3280.2980.3410.3580.3570.3470.3550.3280.3570.3560.3410.3680.3590.368
HVS17.12116.97417.08413.73116.85617.07917.08617.08317.13816.93516.65117.08117.08117.60117.15617.10117.601
HVSm17.20917.07417.18113.78817.07517.18417.18317.18917.25017.03216.75117.17817.17817.70217.24617.19417.702
Img10PSNR23.09522.96323.06122.38222.95222.89023.01922.95824.48022.95122.68123.02223.01322.41623.63423.07524.480
Cielab4.8815.1094.9385.5638.2225.2064.9965.3427.1315.0245.4875.0285.0445.3115.1784.9414.881
SSIM0.5220.5180.5240.4580.4110.4920.5190.5170.5090.5210.4830.5150.5140.4950.5370.5200.537
HVS19.39419.27719.36218.80417.21719.33419.31819.29817.88819.23417.82519.33119.33518.59518.99819.35019.394
HVSm19.68719.61219.68219.19217.67619.69919.65419.65918.12819.55318.09119.66019.66318.85019.24919.66219.699
Img11PSNR24.54424.44324.52420.34722.33624.36724.51324.45522.86624.40623.94624.51324.50121.47124.03124.51524.544
Cielab4.5324.6244.5067.0358.0814.6294.5024.6667.6164.5466.5194.5184.5306.2564.9964.4804.480
SSIM0.5700.5700.5750.4140.4440.5350.5700.5730.5420.5720.4970.5670.5640.4880.5860.5690.586
HVS19.50819.36019.49415.21714.20119.46919.48619.49614.37119.36715.82919.51619.51116.34817.63719.43919.516
HVSm19.79219.65419.79015.37614.34819.79619.78419.80514.46519.65615.97519.81519.81416.50717.80919.72719.815
Img12PSNR21.91221.77621.85018.82419.85321.62021.78521.75920.63021.77420.33821.81521.80421.22421.57321.85121.912
Cielab5.7275.9205.7758.5049.9405.9545.8096.0059.1285.8438.5345.8245.8256.2846.5495.7805.727
SSIM0.6700.6640.6710.5360.5220.6370.6670.6660.6390.6680.5530.6640.6640.6450.6850.6680.685
HVS17.33117.15017.29014.09614.13117.30417.23017.27714.44317.20314.93317.23917.26516.56216.37717.25117.331
HVSm17.54617.39417.52814.25814.37717.58917.47817.54114.57417.44115.09717.48417.51016.76816.54417.48017.589
AveragePSNR23.65023.55023.62321.37722.80023.55323.59123.51023.74323.50123.50123.60323.60422.96124.05923.67624.059
Cielab4.9355.2575.0296.6978.2075.3495.1365.4987.1075.1347.0945.1985.2125.4045.1305.0244.935
SSIM0.5510.5470.5540.4600.4470.5290.5500.5460.5360.5500.4780.5480.5470.5210.5740.5530.574
HVS19.14218.97319.10316.80216.19519.10519.06019.00216.73618.95817.20319.06819.07018.42918.38019.10019.142
HVSm19.43919.29019.41717.07516.54719.44419.38319.34816.95419.26717.45819.38719.38918.69818.60819.40419.444

Appendix C. Performance Metrics of CFA 2.0 at 10 dBs. Three Cases: No Denoising, Denoising After Demosaicing, and Denoising Before Demosaicing

Table A7. Performance metrics of 15 algorithms for CFA 2.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A7. Performance metrics of 15 algorithms for CFA 2.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsBaselineStandardDemonet + GFPCAGSAHCMSFIMPCAGFPCAGLPHPMGSPRACSLSLCDF3ATMFBest Score
Img1PSNR21.32721.37116.72218.57617.3159.87618.82321.46317.2149.87718.80220.00217.33821.89319.93821.893
Cielab7.4529.95212.5039.70211.12330.6219.2736.96111.27430.6209.3408.40012.5017.4098.1756.961
SSIM0.3370.2990.2460.2910.2830.1470.2920.3420.2730.1490.2920.3270.2300.3490.3250.349
HVS15.76315.95611.15613.03311.7544.28513.23315.97511.6674.28513.28614.41411.42516.15114.32016.151
HVSm15.92116.13511.21613.12711.8284.29913.33116.11811.7394.29913.38614.53111.49516.30614.43116.306
Img2PSNR20.95717.29115.99615.01513.41312.65415.41421.45314.16112.19115.15517.91316.10319.89117.11021.453
Cielab6.4638.84610.99311.97214.65616.16411.1915.24913.32417.21411.5398.69310.6496.4929.0445.249
SSIM0.4150.5360.4580.5160.4760.4520.5220.5100.4980.4380.5160.5420.5260.5110.5280.542
HVS16.71612.31911.40210.3348.7147.95510.72816.8999.4767.48510.46013.29711.19015.21612.40216.899
HVSm17.16612.44911.50010.4128.7718.00410.81217.2749.5427.53010.54213.44811.28215.47512.52817.274
Img3PSNR23.89923.61119.15820.01119.14010.06020.72121.04019.45318.76920.69420.71620.43723.19621.34423.899
Cielab6.4677.47010.1508.6449.58930.7418.0077.2539.1649.8218.0358.1718.7496.5367.4586.467
SSIM0.4750.4390.3930.4370.4270.1680.4390.4600.4340.4300.4390.4540.3930.4780.4580.478
HVS18.76918.23414.12614.88414.0084.84915.62515.92714.31613.61915.60115.55215.17017.83316.15718.769
HVSm19.17618.62114.26815.06014.1554.87315.83616.10914.47313.75215.81215.74815.36818.13016.37419.176
Img4PSNR17.35217.56614.40615.73613.77913.66015.79219.05615.42712.42415.99017.02516.29918.18616.84219.056
Cielab10.5869.22414.24012.51014.80714.91611.8817.03412.96217.08211.71311.17310.9888.37710.3617.034
SSIM0.4670.5730.4790.5670.5230.5160.5630.5740.5580.4780.5660.5790.5540.5640.5730.579
HVS13.11912.6959.99311.1319.1399.04011.16214.55810.8517.77411.35512.45111.79313.61012.20614.558
HVSm13.70013.13410.22511.4119.3299.22811.45715.20911.1187.92211.66412.83312.11214.15712.57815.209
Img5PSNR23.46024.75517.50519.15215.6159.97419.89325.27419.05114.57819.85321.93919.06625.08021.85025.274
Cielab5.1785.0529.4977.59811.25524.1786.9703.7817.69912.7907.0045.8977.7554.2805.6533.781
SSIM0.3090.3430.2980.3370.3210.2150.3380.3710.3330.3130.3380.3460.3090.3610.3520.371
HVS19.19420.08613.35414.93611.4005.76715.61720.92214.84710.36915.59217.68514.97420.63917.54320.922
HVSm19.61520.60313.45415.08511.4675.79015.79321.38514.99910.42315.76917.95815.13521.13517.79021.385
Img6PSNR22.21323.05619.42220.42617.39518.38320.68222.96519.94410.27021.10021.59520.40823.69221.87523.692
Cielab8.0798.95110.9039.27112.17210.9078.9285.9319.68433.0568.7078.5399.0356.9247.7395.931
SSIM0.3940.4530.3910.4620.4300.4400.4620.4660.4560.1060.4630.4690.4270.4690.4750.475
HVS18.02019.16715.21516.13413.06114.09516.44618.55215.6655.87016.86617.27416.16919.41017.54419.410
HVSm18.60219.84915.46016.45513.21814.29316.80119.06915.9595.91417.25417.69516.49320.09017.97720.090
Img7PSNR22.10322.30218.04119.60319.14418.82420.21422.99419.43218.98020.09820.50418.24522.67720.95622.994
Cielab6.8156.21710.4218.4458.8939.1207.8335.0178.5978.9827.9157.8009.3445.7147.0855.017
SSIM0.3520.4350.3460.4260.4200.4140.4280.4160.4210.4180.4280.4290.3950.4290.4320.435
HVS17.79417.92713.78815.29714.84814.52715.90318.70115.12314.67215.78316.17713.99118.32216.61218.701
HVSm18.09218.21913.89315.45914.99014.66516.08818.99715.28614.81715.96416.37214.10518.60616.81518.997
Img8PSNR20.85721.74517.62317.68515.65715.73618.36221.73217.58614.88817.98319.65619.42721.87719.60921.877
Cielab7.1107.10410.3419.80712.41112.2128.9125.5509.96413.5329.2708.1238.1546.0497.6435.550
SSIM0.4040.4650.4020.4620.4330.4350.4640.4660.4590.4230.4620.4670.4540.4710.4730.473
HVS16.20216.41412.75212.75210.70710.80013.43116.86212.6739.93413.04414.77614.21616.79914.61216.862
HVSm16.61816.78512.89612.89510.79910.89413.60117.24112.81610.01413.20215.00714.41517.19814.83417.241
Img9PSNR16.52316.38914.22515.03313.53510.08615.00617.47313.58810.08815.05216.22314.18517.31215.98317.473
Cielab7.84810.36810.3209.08810.80616.7818.9706.70410.74316.7968.9218.02410.3827.4518.0096.704
SSIM0.2670.2840.2900.3200.3080.2670.3210.3180.3100.2680.3210.3210.3030.3100.3230.323
HVS12.05112.0249.76610.5409.0335.57210.52112.9839.0885.57310.56711.7359.52712.76111.46812.983
HVSm12.14012.1179.81010.5959.0745.59210.57713.0619.1295.59410.62411.8089.57412.85311.53413.061
Img10PSNR18.22921.92617.84018.90016.36210.02619.17319.94718.16016.15619.45219.49817.28720.17919.51121.926
Cielab8.9976.4769.9428.55811.19126.2208.1296.8079.28511.4067.9368.0879.8856.8757.6766.476
SSIM0.3560.4380.3930.4390.4140.2170.4400.4420.4330.4140.4400.4380.4140.4350.4420.442
HVS14.31718.14813.98914.92112.3986.05415.22615.85814.20712.20615.48715.50813.40216.23515.51318.148
HVSm14.57918.67514.16315.14712.5276.10015.48016.13714.40112.33015.75415.77413.56916.56715.77718.675
Img11PSNR21.30222.12214.88418.28015.76610.05718.74118.45717.08010.05818.76419.32316.77120.92419.36322.122
Cielab7.3077.60714.7329.73112.89528.7259.1098.89011.08328.7269.0848.74911.7257.1898.4497.189
SSIM0.4280.4860.3640.4760.4410.2090.4790.4470.4650.2110.4790.4850.4260.4830.4870.487
HVS16.40616.9259.68913.15110.5904.83513.65613.30411.9284.83613.67914.20911.35315.64214.21416.925
HVSm16.68517.2159.74513.27010.6564.85713.79013.42612.0194.85813.81414.36011.43515.84814.36117.215
Img12PSNR18.44720.65416.77417.57216.27816.22417.98120.28617.15816.05717.98318.26317.38220.08118.52320.654
Cielab8.8106.32011.27710.02811.65611.6909.2207.07010.55011.9409.2219.27210.0876.6728.5886.320
SSIM0.4660.5440.4620.5320.5080.5080.5340.5470.5270.5080.5340.5390.5000.5460.5440.547
HVS14.21016.19212.30313.03011.70011.65313.47416.18912.61011.47313.47613.77612.67915.74714.06016.192
HVSm14.47816.54112.43313.18711.81711.77113.65216.48812.75511.58613.65413.96612.81916.05314.26016.541
AveragePSNR20.55621.06616.88317.99916.11612.96318.40021.01217.35513.69518.41119.38817.74621.24919.40921.249
Cielab7.5937.79911.2779.61311.78819.3569.0356.35410.36117.6649.0578.4119.9386.6647.9906.354
SSIM0.3890.4410.3770.4390.4150.3320.4400.4470.4310.3460.4400.4500.4110.4500.4510.451
HVS16.04716.34112.29413.34511.4468.28613.75216.39412.7049.00813.76614.73812.99116.53114.72116.531
HVSm16.39816.69512.42213.50911.5538.36413.93516.71012.8539.08613.95314.95813.15016.86814.93816.868
Table A8. Performance metrics of 15 algorithms for CFA 2.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A8. Performance metrics of 15 algorithms for CFA 2.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsBaselineStandardDemonet + GFPCAGSAHCMSFIMPCAGFPCAGLPHPMGSPRACSLSLCDF3ATMFBest Score
Img1PSNR21.85120.54723.76520.49720.0759.84020.39024.59919.8249.82620.46920.86722.36523.64422.42624.599
Cielab6.7487.6325.5247.6727.99030.6757.5995.2148.26630.7567.5967.3978.1426.0656.2595.214
SSIM0.3980.4090.4110.4080.4050.1370.4050.4150.4100.1370.4060.4020.4210.4260.4260.426
HVS16.35114.95218.31314.89714.4854.24614.84319.23114.2124.23114.94415.29415.93117.79216.76719.231
HVSm16.46915.03518.46714.97814.5584.26014.92419.42714.2794.24515.02615.38316.04017.93416.88119.427
Img2PSNR15.73019.57321.42918.56715.45415.09718.05121.69118.59915.28218.31918.72821.91821.74320.23521.918
Cielab10.7397.1615.4167.92911.18711.7128.2185.2457.94411.4577.9867.8055.6535.1976.1885.197
SSIM0.3220.4790.3710.4670.4200.4180.4540.3780.4770.4260.4560.4450.4940.4250.4290.494
HVS11.15114.98417.36913.97810.78810.42613.50317.56714.01710.61513.76414.19116.83017.31315.81017.567
HVSm11.26215.20117.73614.14410.87910.50613.65417.96714.17910.69713.92514.36817.15317.67416.06317.967
Img3PSNR26.07826.72823.96926.79025.19218.46226.97426.12225.37424.82027.00826.61328.78026.43826.64428.780
Cielab5.2625.0235.7164.9905.5289.7994.9644.7005.4625.6874.9645.0854.7034.7424.6584.658
SSIM0.5170.5450.5200.5460.5410.5000.5390.5360.5470.5440.5390.5360.5670.5540.5480.567
HVS21.16721.51418.88521.58620.05413.22122.03021.26920.12619.59322.05221.53722.96321.13721.67022.963
HVSm21.67322.02419.16222.09720.39713.31022.60321.73120.48419.90022.63022.04623.67121.57622.16623.671
Img4PSNR14.25216.94520.17217.36114.77514.80017.33419.93617.01314.20417.37917.04719.85020.18118.83520.181
Cielab14.13711.7817.47911.54813.70613.76811.1247.67211.94214.50711.09511.5027.6456.8778.5776.877
SSIM0.4460.6070.5820.6110.5620.5600.6020.5920.6050.5450.6020.5910.6590.6360.6180.659
HVS9.78412.29716.36212.72910.12010.13512.83315.95712.3989.53212.86612.45015.16915.97014.50816.362
HVSm10.00912.61317.06713.07010.33010.34713.18916.60712.7179.72213.22612.78215.68416.57314.97617.067
Img5PSNR22.83024.50920.60523.17620.24810.54224.92025.69222.70220.93524.81524.57326.02923.91125.07826.029
Cielab5.1954.5076.0705.0236.58022.0414.3563.6375.2716.1904.3844.4984.0024.3043.9613.637
SSIM0.3040.3490.3030.3450.3360.1880.3430.3190.3490.3430.3420.3390.3540.3330.3430.354
HVS18.69820.20116.40118.90216.0036.33020.73921.59018.43816.68520.60520.33421.58119.68120.90021.590
HVSm18.96220.50916.53919.12916.1326.35521.08222.02018.63716.82120.94320.65422.01619.95621.24922.020
Img6PSNR22.65122.13621.99621.80621.65620.94722.11523.82721.03510.43322.21122.12924.74823.59122.97824.748
Cielab7.3697.5236.7747.6727.7738.1757.7135.6058.19131.6907.6677.5425.8735.6846.3425.605
SSIM0.3450.4050.3080.4020.3990.3990.3920.3260.4060.0680.3910.3840.4120.3590.3710.412
HVS18.61517.81317.73417.44817.39616.63117.93119.85016.6756.03118.04317.84120.04819.30818.82520.048
HVSm19.16218.24118.14217.85917.76316.95218.39120.52617.0156.07918.51418.29220.76119.89019.36120.761
Img7PSNR25.44927.12226.66927.07226.75326.65426.86428.64426.73026.62226.97426.55429.73428.78728.14329.734
Cielab5.2184.6713.7824.6864.7204.7874.7053.1994.7924.7944.6834.8613.0952.9243.6182.924
SSIM0.3400.4300.3520.4290.4290.4310.4190.3540.4350.4340.4190.4090.4680.4090.4120.468
HVS21.47122.79323.04422.76822.50522.38622.75325.67622.43022.36622.83922.35925.83825.40524.46725.838
HVSm22.05823.52323.77423.48523.16123.02023.47327.15023.08322.99023.57623.01727.27426.69625.50427.274
Img8PSNR19.78822.45726.11122.26720.56120.09221.64923.79722.03916.38521.65122.74526.63625.70823.64326.636
Cielab7.7276.1473.9006.2377.2177.5356.4524.7576.40611.1246.4636.0414.2163.9694.9423.900
SSIM0.3880.4720.4000.4700.4570.4590.4530.3970.4770.4130.4530.4540.4840.4390.4440.484
HVS15.13217.54722.35617.38915.67615.18016.91219.57117.16811.41616.89118.00120.98521.11919.04522.356
HVSm15.36917.88123.27617.69715.88815.36817.19920.08717.45711.51217.17918.36421.68321.81319.48423.276
Img9PSNR15.22117.63721.88818.08215.28510.08818.03820.03018.1239.97317.91218.29424.05621.90619.58124.056
Cielab8.8906.8464.4846.5568.73916.6346.5245.2556.59616.9036.6006.4484.3824.5215.4564.382
SSIM0.2480.2990.2820.3010.2850.2260.2990.2800.3050.2250.2990.2910.3060.2950.2890.306
HVS10.74813.14517.58113.59210.7865.56813.59615.63613.6345.45113.46813.81618.99417.35115.14718.994
HVSm10.79713.21317.73713.66610.8295.58713.67015.74213.7055.47013.54013.89419.22417.50115.24319.224
Img10PSNR18.84419.63820.69620.10119.04510.16920.05121.77919.18318.83820.06219.81221.27621.31820.81921.779
Cielab8.4297.7936.5397.4708.23325.4407.4105.7268.2368.4227.4047.6556.3385.9236.4075.726
SSIM0.3410.4090.3550.4130.4030.1690.4030.3610.4150.4100.4020.3920.4080.3820.3920.415
HVS14.98915.64116.83716.07215.0846.19216.13618.02115.16914.86116.15415.84117.32117.47116.95718.021
HVSm15.24115.90017.13316.37015.3026.24216.44218.44915.40915.07016.46016.12717.70517.83417.29818.449
Img11PSNR17.73320.63917.75120.58716.90410.11820.71419.41720.47710.10120.72020.59321.20019.39720.08721.200
Cielab9.9747.4589.8127.49410.96928.2417.3137.9587.61828.3267.3087.4887.3188.1147.4667.308
SSIM0.3260.4110.2810.4110.3620.1320.4010.2980.4240.1320.4010.3910.3900.3320.3670.424
HVS12.64815.56812.79815.51511.7504.89615.75914.42215.3934.87915.76415.56715.54414.22315.04715.764
HVSm12.75515.75412.90315.69811.8314.92015.95414.57515.5684.90315.96015.75415.74114.36615.21515.960
Img12PSNR17.84518.76918.57718.85317.88717.78619.12818.18318.27917.86619.14118.93122.00519.52218.94322.005
Cielab9.3658.4979.2358.4239.3419.4857.9409.6719.0219.4107.9318.3555.7507.8728.3295.750
SSIM0.4240.5070.4510.5080.4920.4960.5000.4540.5080.5000.5000.4960.5430.4930.4880.543
HVS13.55914.29014.47214.38213.38913.26214.77314.08013.77213.34314.78614.53117.48015.30414.69117.480
HVSm13.75314.49914.65514.59513.55713.42815.00714.25013.95713.51015.02014.75217.86815.52414.89817.868
AveragePSNR19.85621.39221.96921.26319.48615.38321.35222.81020.78116.27421.38821.40724.05023.01222.28424.050
Cielab8.2547.0876.2287.1428.49915.6917.0275.7207.47914.9397.0077.0565.5935.5166.0175.516
SSIM0.3670.4430.3850.4430.4240.3430.4340.3930.4470.3480.4340.4270.4590.4230.4270.459
HVS15.36016.72917.67916.60514.83610.70616.81718.57216.11911.58416.84816.81319.05718.50617.82019.057
HVSm15.62617.03318.04916.89915.05210.85817.13219.04416.37411.74317.16717.11919.56818.94518.19519.568
Table A9. Performance metrics of 15 algorithms for CFA 2.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A9. Performance metrics of 15 algorithms for CFA 2.0 pattern at 10 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsBaselineStandardDemonet + GFPCAGSAHCMSFIMPCAGFPCAGLPHPMGSPRACSLSLCDF3ATMFBest Score
Img1PSNR23.14023.35129.93623.35023.32322.23623.19623.01623.38122.13523.20623.28519.63825.89423.37429.936
Cielab5.4285.3873.6485.3895.4385.4965.3115.9045.4025.5025.3815.39710.8814.3785.3793.648
SSIM0.4270.4440.4130.4440.4380.4440.4410.4200.4450.4430.4420.4370.3370.4400.4410.445
HVS17.80317.88525.07017.88717.87517.88917.62817.55817.89717.89517.76717.87113.11620.54017.91925.070
HVSm17.97818.02326.01218.02518.01518.03117.76017.70518.03918.03717.90318.01913.18320.79518.05726.012
Img2PSNR23.27524.46325.60824.45124.43224.39624.20723.96424.41024.40624.24924.28819.79925.21124.44025.608
Cielab4.3684.0093.4894.0374.0784.0854.0624.0714.0654.0894.0504.1068.9963.6834.0323.489
SSIM0.3710.5010.3650.5000.4990.4990.4970.4490.5000.4990.4960.4820.4850.4660.4950.501
HVS19.60120.44421.74220.50820.46920.52720.21619.81520.52820.51920.23920.41314.25321.36820.50221.742
HVSm20.48321.14423.25821.18621.15421.21620.85120.57321.22621.21520.88721.13014.42522.35121.20223.258
Img3PSNR29.21930.27227.45730.26430.14829.94730.21130.02229.95629.92230.21230.02824.98329.61630.19430.272
Cielab4.2154.0924.6254.0964.2064.0344.1264.2294.0964.0384.1274.1499.2934.1214.1014.034
SSIM0.5530.5810.5330.5810.5770.5770.5780.5740.5770.5760.5780.5740.5340.5730.5780.581
HVS25.35225.67622.76525.66825.65125.07725.76325.57324.94024.99925.76525.64618.07925.08325.67125.765
HVSm26.82726.96923.50726.95726.94926.36027.04026.76626.21826.27827.04326.96418.30326.15826.95227.043
Img4PSNR18.90220.44820.76920.44520.42120.40020.00719.80820.36720.41920.01620.29818.25620.82620.42720.826
Cielab8.1877.6095.9787.6087.6077.5747.5976.5337.7887.5837.5977.5989.1166.5677.4545.978
SSIM0.4960.6590.5560.6590.6560.6490.6480.6120.6500.6490.6480.6440.6220.6450.6540.659
HVS15.00915.89216.48015.90415.90415.96015.45215.29215.91815.95615.44615.81913.38216.40815.91916.480
HVSm15.82316.47417.45016.48316.48316.57116.01015.94816.52716.57416.00816.43513.76017.10916.51617.450
Img5PSNR25.74126.54129.48426.53926.51026.52626.25426.14226.52926.52926.28626.42124.70628.49926.55929.484
Cielab4.1113.9843.1703.9773.9813.9774.0694.1653.9763.9794.0403.9835.7843.4483.9483.170
SSIM0.3110.3590.3090.3590.3570.3590.3560.3420.3570.3570.3550.3510.3430.3470.3570.359
HVS22.24622.88425.18622.90722.88022.96622.42722.24022.97622.96822.51722.82820.68024.90722.93325.186
HVSm22.97123.40226.50223.41823.39823.47822.90422.79523.49323.48623.01023.37921.04225.84023.45926.502
Img6PSNR23.08823.72325.58323.72423.70323.73123.50823.44123.73723.73723.51223.61522.75425.85423.73425.854
Cielab6.1325.8755.4835.8755.9465.8855.9645.8925.9045.8965.9435.9558.9905.0515.8505.051
SSIM0.3430.4190.3340.4190.4160.4200.4150.3950.4210.4200.4130.4040.3920.3990.4160.421
HVS19.33919.79321.15419.75419.82119.85419.64019.34919.82319.86119.56019.67717.53721.91319.81921.913
HVSm20.04920.38222.26820.35820.39620.45120.23919.94720.42820.46020.14920.31017.89623.01320.42823.013
Img7PSNR27.57328.66326.64228.67028.65128.56728.71328.44228.54928.53828.70428.55127.22128.14028.63428.713
Cielab4.0933.9134.2823.9173.9323.9403.9013.8233.9403.9453.8953.9466.2453.9553.9213.823
SSIM0.3830.4710.3520.4710.4700.4700.4680.4390.4700.4700.4680.4610.4720.4420.4670.472
HVS24.41725.30023.04525.31025.32325.24825.40725.04425.21025.20025.33525.22123.14124.77425.29325.407
HVSm25.70526.50323.98526.51026.52526.46426.64326.29826.42626.41226.55026.44123.81525.90626.50126.643
Img8PSNR25.70028.63825.48728.63128.54128.30528.01927.10028.25428.11028.06628.15323.48729.04428.59029.044
Cielab3.8183.4303.9793.4353.5123.4833.5323.5333.4863.4993.5383.5466.8293.1963.4313.196
SSIM0.4190.5000.4020.5000.4980.4990.4920.4660.4990.4970.4930.4870.4880.4770.4960.500
HVS22.58725.01421.18725.08525.11524.80424.41223.38024.68124.56024.40724.74317.23225.59325.15625.593
HVSm24.27226.53522.11126.57326.61226.41025.71524.85326.28326.17825.72326.34717.51827.57626.72227.576
Img9PSNR24.64125.35028.28725.35925.32423.29425.12725.08525.34323.05725.12225.28120.68526.47125.35428.287
Cielab3.6223.4392.8923.4373.4613.6273.4853.5653.4583.7163.4873.4506.5563.1513.4252.892
SSIM0.2620.3080.2690.3080.3070.3030.3080.2970.3030.2990.3080.3010.3350.2980.3060.335
HVS20.75621.18524.48821.19021.19821.23220.97020.89821.22221.23520.96121.13915.25422.42221.20724.488
HVSm21.27321.50725.49921.51621.51021.55721.28421.25621.54821.56321.27521.49215.35722.89021.53525.499
Img10PSNR21.01921.55923.82821.55721.54621.58821.37321.34421.58821.59321.36921.45221.13622.41821.56223.828
Cielab6.4246.2034.8596.2046.2456.2176.2196.1756.2586.2216.2226.2577.7175.5966.1794.859
SSIM0.3440.4160.3290.4160.4140.4200.4130.3920.4200.4190.4100.4010.4230.3940.4120.423
HVS17.57317.99920.17217.95918.02818.09917.82017.69918.05618.10517.76717.88517.06918.85518.00520.172
HVSm18.09818.39021.15118.36718.40518.47918.23018.12818.45018.48618.17318.32317.38919.40218.41621.151
Img11PSNR22.20822.58726.80322.58722.56122.57622.44922.37622.60922.50322.44522.50821.88924.40022.59726.803
Cielab5.8715.7634.4555.7685.8045.7905.7605.8905.7745.7915.7585.7788.2374.9675.7474.455
SSIM0.3450.4010.3140.4010.3980.4070.3970.3790.4090.4080.3960.3870.3920.3790.3980.409
HVS17.82018.01522.73218.01518.02618.05117.88217.79418.05018.05217.87517.98015.77520.05318.04722.732
HVSm18.18718.32324.01718.32318.32918.35818.18418.11918.35818.36118.17718.30315.97120.58818.35724.017
Img12PSNR21.00521.72823.81021.72121.69421.71321.51821.49321.72621.73021.51921.62418.44722.61021.72623.810
Cielab6.3886.2224.8536.2276.2516.2266.1496.0246.2576.2266.1506.23711.0585.5586.1964.853
SSIM0.4570.5430.4480.5430.5410.5440.5360.5190.5450.5440.5360.5320.5300.5190.5390.545
HVS17.54517.85920.61617.85217.86517.88017.63517.70417.87617.89017.63617.82812.90918.96917.88920.616
HVSm18.10318.29821.69718.29018.29818.33118.06118.14918.32718.34218.06218.28413.03619.55618.33121.697
AveragePSNR23.79324.77726.14124.77524.73824.44024.54824.35324.70424.39024.55924.62521.91725.74924.76626.141
Cielab5.2214.9944.3104.9975.0385.0285.0144.9845.0345.0405.0165.0338.3094.4734.9724.310
SSIM0.3930.4670.3850.4670.4640.4660.4620.4400.4660.4650.4620.4550.4460.4480.4630.467
HVS20.00420.66222.05320.67020.67920.63220.43820.19620.59820.60320.44020.58716.53621.74120.69622.053
HVSm20.81421.32923.12121.33421.33921.30921.07720.87821.27721.28321.08021.28616.80822.59921.37323.121

Appendix D. Performance Metrics of CFA 2.0 at 20 dBs. Three Cases: No Denoising, Denoising After Demosaicing, and Denoising Before Demosaicing

Table A10. Performance metrics of 15 algorithms for CFA 2.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A10. Performance metrics of 15 algorithms for CFA 2.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. No denoising. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsBaselineStandardDemonet + GFPCAGSAHCMSFIMPCAGFPCAGLPHPMGSPRACSLSLCDF3ATMFBest Score
Img1PSNR22.12022.85820.03520.24420.21620.14520.13722.57820.22820.15620.19420.88519.71423.14921.33523.149
Cielab6.2638.7938.3577.6637.7367.9607.5866.1467.7427.9637.5937.1279.5426.3636.8666.146
SSIM0.4350.4160.3580.4110.4020.3830.4120.4430.3910.3860.4120.4410.3310.4590.4550.459
HVS16.68417.62014.56714.71914.70114.71014.53617.10914.71914.71014.69115.35913.66017.44815.70217.620
HVSm16.83017.79714.65514.80414.79314.80314.61817.25414.81014.80414.77715.45513.74217.60315.80217.797
Img2PSNR21.84920.77020.14920.29520.27520.25620.07123.06920.26820.27420.09920.25020.41522.25021.08023.069
Cielab5.6616.5767.0416.6606.7566.7196.6134.3946.7076.7166.6006.6976.4895.0685.7284.394
SSIM0.4460.5960.5520.6190.6150.6100.6170.5600.6140.6130.6160.6060.6250.5580.6160.625
HVS17.73815.57315.79215.69415.68315.70215.45518.51115.69015.69815.46115.67015.42417.54116.37018.511
HVSm18.26915.81215.99315.89015.88015.90415.64019.00915.89615.90315.65215.87815.59517.94616.61419.009
Img3PSNR25.68126.43620.25822.66620.55920.44223.42521.57520.97820.45523.43224.30124.49625.43124.36926.436
Cielab4.9516.1808.2996.2817.8637.8995.8006.7727.4257.9005.8075.5375.8295.2745.3374.951
SSIM0.5500.5300.4870.5370.5220.5230.5380.5460.5290.5260.5380.5500.4910.5650.5680.568
HVS20.81521.86015.19417.52215.40215.27618.31816.39515.80715.28118.33619.20119.16120.03819.11221.860
HVSm21.30822.46115.31417.71315.52515.39118.55416.55215.93715.39618.57419.48219.46020.40019.36922.461
Img4PSNR17.97718.31418.35118.93218.88918.89618.50620.31818.88718.93618.51518.91518.73819.08819.11620.318
Cielab10.0008.73810.6089.9779.8859.7849.7596.35410.1309.7699.7699.7888.8797.7968.4456.354
SSIM0.4820.5980.5420.6150.6110.6050.6080.6010.6090.6090.6080.6120.5890.5880.6230.623
HVS13.82213.43714.59614.41614.40314.47613.92515.89414.42914.46913.91514.37114.53014.55114.54015.894
HVSm14.49113.92515.16514.95514.95015.04814.43016.72614.99615.04414.42114.92215.06815.19515.10016.726
Img5PSNR23.95526.08320.33920.44420.42820.43121.03026.18720.43720.44120.71122.97422.21226.48623.35226.486
Cielab4.5494.4106.8576.3966.4276.4205.9563.4106.4296.4206.1525.0305.4793.6714.6063.410
SSIM0.3490.4060.3670.4030.4000.3950.4030.4330.3970.3970.4030.4090.3740.4200.4370.437
HVS19.83021.24516.26116.22516.21916.24416.74221.79316.23916.24316.44918.74618.13622.02719.08422.027
HVSm20.24621.77516.40016.36516.35816.38316.90222.25816.38216.38516.60218.99118.37022.59719.32922.597
Img6PSNR24.14223.85720.52123.50320.39720.65623.70226.23522.28020.44024.03623.80021.88026.35024.78826.350
Cielab5.8468.1688.5346.2738.3438.0326.1384.3077.0088.2186.0036.1727.1865.4735.2634.307
SSIM0.4850.5940.5480.6320.6040.6040.6310.6030.6210.6050.6310.6210.6080.5930.6360.636
HVS20.41220.22616.30119.22116.10016.39119.52821.75518.02416.16219.83219.54217.32822.37820.48522.378
HVSm21.25220.83016.51319.63716.30316.60819.99622.59218.34216.36920.32720.03417.58823.44821.06823.448
Img7PSNR26.19226.24321.91123.30423.00922.48224.01225.39923.01022.32523.87124.28522.94926.34124.82826.341
Cielab4.4054.1876.5875.5245.7225.9625.1053.8455.6836.0535.1685.1175.3463.8714.4823.845
SSIM0.4410.5570.4780.5540.5490.5410.5540.5190.5470.5440.5540.5530.5380.5390.5650.565
HVS22.31221.92317.85619.01518.73818.21919.72421.16718.72718.04119.57720.00618.76022.15220.54522.312
HVSm23.02322.39818.02919.25018.95618.42220.00221.59118.96218.23819.84820.30918.96422.67720.86523.023
Img8PSNR22.49222.68721.20620.28620.25820.27319.99624.89520.27820.28020.01121.18721.04424.03722.02924.895
Cielab5.6806.5266.9357.2397.4017.2907.2323.9557.2987.2927.2316.6866.6564.8315.7013.955
SSIM0.4720.5450.5090.5630.5560.5550.5590.5540.5590.5560.5590.5560.5530.5510.5860.586
HVS18.08917.43816.64315.37915.39115.43615.06820.18315.41215.43015.08416.33915.74619.02916.94220.183
HVSm18.67117.84616.88915.56715.57415.62415.25420.87615.60715.62215.27516.59515.95019.61717.23520.876
Img9PSNR20.40117.18720.50720.53920.51320.43920.37820.61720.51620.44820.37520.54720.34219.99720.47720.617
Cielab5.4129.8195.7215.2635.3565.7095.2544.9045.3365.7295.2535.2685.5376.0715.0264.904
SSIM0.2890.3030.3300.3530.3480.3400.3530.3440.3480.3410.3530.3500.3410.3310.3630.363
HVS16.03312.90616.28516.12416.11316.13015.97316.15816.12016.12915.97016.11715.71115.49415.97616.285
HVSm16.22913.00816.42916.27916.26916.28816.12516.29616.27916.28916.12216.27715.86115.64316.11416.429
Img10PSNR21.32923.64120.09420.62320.20520.22220.79922.42220.22620.23621.18121.61420.55822.75621.79823.641
Cielab6.3535.4017.6016.9307.2737.1956.6375.1657.2867.1946.4166.3306.8695.2075.7895.165
SSIM0.4140.5090.4730.5120.5050.5030.5120.5140.5060.5050.5120.5090.5000.5040.5280.528
HVS17.66019.94916.35216.68416.31616.37216.91118.37316.33316.37517.26617.70116.69018.94917.86719.949
HVSm18.16020.58216.57316.94416.53816.59817.20218.79316.56816.60217.57718.05116.93319.47218.21620.582
Img11PSNR22.03323.55620.19620.38320.35120.36520.20820.14720.37620.37420.20720.76420.24322.50521.05023.556
Cielab6.2696.8898.0347.4767.5947.5317.4487.3197.5267.5307.4447.1667.8935.9996.7585.999
SSIM0.4890.5840.5250.5920.5830.5840.5890.5430.5890.5880.5890.5860.5620.5670.6040.604
HVS17.29318.69715.21715.28715.28315.30715.14015.04115.29915.30615.13615.70114.77217.26415.86418.697
HVSm17.58419.03815.35215.42215.42015.44415.27415.19515.43815.44515.27115.85714.89817.51316.02119.038
Img12PSNR19.71920.47720.20520.36120.33120.34920.12022.65520.36320.37320.12320.30220.53221.09120.66122.655
Cielab7.2575.9937.4557.1477.2177.1576.9824.8817.1907.1556.9837.1446.9935.6436.3894.881
SSIM0.5280.6250.5890.6450.6400.6380.6400.6350.6420.6410.6400.6400.6270.6210.6540.654
HVS15.73415.93816.03615.98515.99116.03115.74818.73416.01416.03615.75015.97015.82116.88316.28818.734
HVSm16.08316.22316.25516.22416.22916.27615.98519.22116.26116.28315.98816.22216.02217.23916.55219.221
AveragePSNR22.32422.67620.31420.96520.45220.41321.03223.00820.65420.39521.06321.65221.09423.29022.07423.290
Cielab6.0546.8077.6696.9037.2987.3056.7095.1217.1477.3286.7026.5056.8925.4395.8665.121
SSIM0.4480.5220.4800.5360.5280.5230.5350.5250.5290.5260.5350.5360.5120.5250.5530.553
HVS18.03518.06815.92516.35615.86215.85816.42218.42616.06815.82316.45617.06016.31218.64617.39818.646
HVSm18.51218.47516.13116.58816.06616.06616.66518.86416.29016.03216.70317.34016.53819.11217.69019.112
Table A11. Performance metrics of 15 algorithms for CFA 2.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A11. Performance metrics of 15 algorithms for CFA 2.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsBaselineStandardDemonet + GFPCAGSAHCMSFIMPCAGFPCAGLPHPMGSPRACSLSLCDF3ATMFBest Score
Img1PSNR19.95820.01325.37221.90520.74019.85621.95623.12221.31519.29421.98521.55821.85923.28222.33125.372
Cielab8.3988.2254.7106.3047.1248.0716.1435.7446.7608.5866.1816.5217.9215.5236.0574.710
SSIM0.3670.3740.4660.4910.4810.4800.4880.4610.4950.4750.4890.4760.4720.4750.4910.495
HVS14.56014.58119.96116.36615.20014.33016.46117.68315.75313.75516.51416.05515.49717.82816.74519.961
HVSm14.65514.66820.15216.45315.26914.38916.55117.80415.82713.80716.60516.14015.58417.94716.83920.152
Img2PSNR19.72120.11822.56220.64920.50719.99620.43721.59320.51620.13920.65320.15423.85621.44921.20723.856
Cielab8.1107.4474.8076.3726.4736.8326.3795.2846.4966.7386.2506.6844.6045.4455.6634.604
SSIM0.4920.6100.4480.5550.5530.5530.5470.4470.5620.5570.5470.5250.5910.4800.5290.610
HVS15.45515.54518.61616.06315.91115.40015.92317.33315.93915.55016.12415.61818.71517.16316.74318.715
HVSm15.68115.74019.00616.27116.11915.57416.12617.64716.13515.72616.33915.81519.07817.44316.98619.078
Img3PSNR20.07320.16423.41224.80624.16520.79024.57922.80624.46924.16324.59924.52629.20123.59624.42529.201
Cielab8.7768.4695.6805.0805.4557.4695.1505.9835.2805.4925.1545.2554.2375.5095.1424.237
SSIM0.4870.5010.5750.6100.6050.5890.6040.5790.6130.6110.6040.5980.6220.5910.6040.622
HVS15.03115.06318.32319.59619.01215.55119.49617.67219.23318.94219.51319.40123.13918.48019.27823.139
HVSm15.15915.18118.51619.83319.20915.65519.73217.84319.44119.13019.75019.63023.65418.67219.49923.654
Img4PSNR17.93418.58520.13519.16119.04918.93918.86620.22918.96118.85618.84518.96220.44319.90619.57420.443
Cielab13.43012.4338.31211.00610.83811.00410.7938.35211.26811.02910.80710.8627.6478.6399.3397.647
SSIM0.5010.5860.5900.6340.6320.6270.6270.6120.6290.6280.6260.6240.6550.6230.6420.655
HVS14.08714.30116.40114.59014.49914.39614.43216.28214.43314.29714.39314.40915.96315.79115.19316.401
HVSm14.68714.84617.12615.09014.99014.88114.92616.97814.91914.76914.88514.90316.56816.40915.73217.126
Img5PSNR20.07820.18321.41422.63922.02021.65324.39324.51021.99521.15123.65525.29325.51723.64123.99925.517
Cielab7.3987.1095.4985.0215.3315.5684.2513.9985.3895.8664.5253.9654.0334.3534.2523.965
SSIM0.3400.3820.3340.3830.3810.3830.3820.3490.3870.3830.3800.3790.3890.3600.3800.389
HVS15.99916.04117.27818.39917.78317.42120.22420.36517.76216.92119.46621.11121.10519.50719.80221.111
HVSm16.15416.18117.42418.56717.93617.55120.47220.64317.90517.03919.68221.41521.41619.72620.02621.416
Img6PSNR19.92120.21121.68524.06221.95722.21024.04723.04223.52121.81624.22124.57724.27623.09423.87524.577
Cielab9.8079.1126.8085.8126.9946.8076.0375.8536.1687.0585.9635.6485.6515.8985.6075.607
SSIM0.4920.5950.4200.5580.5350.5470.5450.4450.5640.5460.5430.5310.5610.4740.5330.595
HVS15.82315.89317.50119.74117.69717.91719.89818.91319.21017.51420.08720.39919.54318.94919.64620.399
HVSm16.04316.09217.79520.22617.97718.21920.40319.31219.62717.79020.61120.96519.97919.34520.10420.965
Img7PSNR19.93020.07329.06430.26830.45528.68630.46530.20729.17327.92730.51230.35032.02530.15130.72732.025
Cielab8.6228.1822.8453.3773.3363.6783.3332.6663.5863.8363.3293.3832.4762.7732.8902.476
SSIM0.4300.5190.4640.5650.5660.5650.5590.4750.5690.5640.5600.5470.5970.5080.5540.597
HVS15.76015.79826.19026.15326.44924.55126.59627.69325.03523.77026.60826.52428.26727.22427.23428.267
HVSm15.88715.91327.33027.29127.64425.28927.87729.43025.88424.37427.89927.80129.99528.68428.66629.995
Img8PSNR19.64320.06524.84521.58220.24919.78320.92723.51220.63619.39921.10521.39724.92123.26322.39224.921
Cielab8.7128.0424.2866.3047.2227.5646.5984.8626.9617.8756.5006.4574.6025.0245.5024.286
SSIM0.4640.5440.4600.5410.5280.5290.5230.4650.5430.5270.5260.5160.5660.4880.5260.566
HVS15.10515.21120.80816.67915.35514.86916.12819.23115.72614.47916.29816.58419.30518.78417.60320.808
HVSm15.32515.39921.35116.90615.52515.02116.33919.63415.91214.62016.51716.81719.69819.13817.87521.351
Img9PSNR20.24620.34421.54420.41320.14420.13420.03221.56120.28520.36020.12420.19122.62321.10220.78222.623
Cielab6.4686.0384.5675.2175.3575.6985.3444.5545.3445.6145.3005.3414.5894.7194.8574.554
SSIM0.3010.3430.3000.3270.3240.3230.3250.3000.3310.3240.3250.3140.3310.3020.3190.343
HVS16.00016.03617.24615.94315.69315.72115.61617.20915.81815.95415.71115.73717.68816.73116.35117.688
HVSm16.17516.19617.37816.04815.78715.82015.71617.34015.91716.05615.81215.84117.83916.84916.46117.839
Img10PSNR19.72120.02522.21622.35020.77521.18522.30323.54721.74320.21722.50221.54624.58322.45622.64824.583
Cielab8.6648.0305.5225.9116.8146.5705.8654.8436.3457.1995.7606.3314.7115.4075.3634.711
SSIM0.4280.4970.4200.4860.4740.4840.4790.4310.4900.4790.4780.4610.4900.4430.4710.497
HVS16.03616.11918.57018.37216.87017.26818.46419.93717.78816.28918.67417.64820.64518.73718.82620.645
HVSm16.28616.34018.90718.75117.11217.53218.85520.42118.10916.50619.08017.97821.18519.10719.21621.185
Img11PSNR19.93820.15520.31620.82220.35319.90920.97220.48720.75819.81320.90721.19721.39620.68520.87521.396
Cielab8.6328.1867.2196.9917.3567.7186.8007.0337.0907.7986.8436.7346.9506.8726.7526.734
SSIM0.4920.5700.3880.5070.4990.5080.4990.4020.5220.5100.4980.4830.4990.4320.4790.570
HVS15.08515.12415.38115.74415.29514.80515.99415.53515.67114.70715.92416.20315.81415.72715.83516.203
HVSm15.23015.25715.53315.89915.42814.92716.15915.69415.82014.82516.08716.37515.97915.88615.99416.375
Img12PSNR19.74120.21620.02020.20019.78120.05419.80920.26620.17119.99819.82819.96021.98820.11820.17821.988
Cielab8.5538.0147.1137.1467.4707.2847.2206.8927.2277.3367.2077.3655.7917.0076.9235.791
SSIM0.5450.6240.5440.6150.6090.6170.5990.5580.6230.6190.5990.5970.6410.5720.5970.641
HVS15.75415.83215.91415.78915.37615.64715.46716.17915.76015.58115.48815.61517.30815.93715.86517.308
HVSm16.00516.05916.13316.02415.58315.86915.69116.41515.98915.79815.71315.84317.59416.16116.09217.594
AveragePSNR19.74220.01322.71522.40521.68321.10022.39922.90721.96221.09522.41122.47624.39122.72922.75124.391
Cielab8.7988.2745.6146.2126.6477.0226.1605.5056.4937.0366.1526.2125.2685.5975.6965.268
SSIM0.4450.5120.4510.5230.5160.5170.5150.4600.5270.5190.5150.5040.5350.4790.5100.535
HVS15.39115.46218.51617.78617.09516.49017.89218.66917.34416.48017.90017.94219.41618.40518.26019.416
HVSm15.60715.65618.88818.11317.38216.72718.23719.09717.62416.70318.24818.29419.88118.78118.62419.881
Table A12. Performance metrics of 15 algorithms for CFA 2.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
Table A12. Performance metrics of 15 algorithms for CFA 2.0 pattern at 20 dBs SNR (Poisson noise). Bold numbers indicate the best performing method in each row. Denoising is applied after CFA is generated and before CFA is demosaiced. Red numbers indicate those methods used in F3 and red and green numbers indicate those methods used in ATMF.
ImageMetricsBaselineStandardDemonet + GFPCAGSAHCMSFIMPCAGFPCAGLPHPMGSPRACSLSLCDF3ATMFBest Score
Img1PSNR29.72430.92829.45630.93230.77830.72630.50029.50130.93030.71030.65230.54024.15330.62530.93430.934
Cielab2.9272.8353.5112.8322.9212.9542.8103.5682.8502.9682.8412.8676.9133.0092.8522.810
SSIM0.4950.5210.4600.5210.5130.5220.5190.4900.5230.5190.5190.5100.4270.5120.5190.523
HVS25.44525.96324.27226.01025.95525.94925.28324.27925.91225.94525.75325.88517.49325.70726.01526.015
HVSm26.37926.67424.93326.70626.66326.66825.87124.85726.65026.67126.41626.65317.63026.35926.70126.706
Img2PSNR24.94327.24026.72827.22227.18327.07826.88826.25727.09127.07226.93626.89223.68027.03327.17127.240
Cielab3.6723.1823.0503.2053.2623.2693.2493.2443.2463.2773.2443.2986.0493.1313.2033.050
SSIM0.4100.5710.4350.5700.5680.5670.5660.5060.5670.5660.5650.5490.5970.5530.5670.597
HVS21.77723.45822.97123.59023.51423.61323.21022.23723.57423.55523.18623.39117.99223.29223.59323.613
HVSm23.24124.67024.73924.76024.69924.82424.27523.48224.80124.78124.27124.63518.26924.53724.78624.824
Img3PSNR30.68232.96129.66332.96132.70432.66132.64931.93732.67332.63532.66432.40327.45232.81532.91932.961
Cielab3.1162.9153.5972.8983.0692.9222.9733.2192.9212.9292.9782.9936.1762.9452.8802.880
SSIM0.5930.6350.5930.6350.6300.6320.6310.6210.6310.6300.6310.6250.5860.6300.6320.635
HVS27.77629.11625.48029.13429.13028.60328.86627.91028.38728.49628.91728.91521.01729.15529.07029.155
HVSm30.27631.26826.50431.27631.26230.68330.85129.55730.44330.56430.93531.15621.33331.21931.18731.276
Img4PSNR19.96722.07620.46822.07522.01721.94821.53821.36321.90821.96821.54921.88919.85222.04622.02122.076
Cielab8.1817.6206.0377.6177.6027.5377.5845.7687.8127.5437.5907.5778.1456.5597.3415.768
SSIM0.5060.6730.5770.6720.6680.6620.6640.6340.6620.6620.6640.6590.6270.6700.6690.673
HVS16.29917.49016.16117.50417.50017.53616.95316.91517.45817.51316.92817.39015.36917.62817.53017.628
HVSm17.44218.37717.03618.38818.39018.48117.77217.87018.40018.46717.74918.31615.93318.55418.43418.554
Img5PSNR28.68830.63530.20330.63430.55730.52330.10529.73830.52130.50830.20330.33425.78530.45930.60130.635
Cielab2.6832.5042.5272.4702.5092.4982.6152.7102.4962.5032.5702.5095.0332.5082.4742.470
SSIM0.3370.3950.3430.3950.3930.3940.3910.3750.3920.3920.3910.3860.3840.3900.3930.395
HVS25.53627.06326.12327.11127.05727.12926.23725.61427.12927.09826.41226.91621.74826.80027.13627.136
HVSm27.05928.18627.55728.21528.17828.28327.17026.67628.29428.27527.40628.11822.10027.92228.23628.294
Img6PSNR26.19728.32728.08928.32928.23128.25027.91727.45128.26428.22927.94727.92625.40328.14028.27728.329
Cielab4.3173.8583.9983.8313.9713.8804.0554.0013.9073.9004.0343.9915.9653.8183.8433.818
SSIM0.4310.5630.4590.5630.5590.5620.5560.5190.5630.5610.5550.5370.5680.5510.5600.568
HVS23.10924.62823.99824.50124.69924.75224.18923.26924.65524.73024.12924.19620.40024.42924.65924.752
HVSm24.76726.01025.59525.92726.04626.17525.56124.49026.11026.16525.47625.71220.87025.76626.04926.175
Img7PSNR27.61829.20426.93829.23029.20529.10529.35028.80429.07429.05329.34329.04530.46229.20229.20330.462
Cielab3.3833.0763.5133.0833.1143.1163.0623.0343.1113.1253.0563.1354.4773.0093.0773.009
SSIM0.4470.5800.4550.5790.5770.5770.5760.5300.5760.5750.5760.5650.5970.5660.5750.597
HVS24.57225.73123.40825.78825.80625.72025.99125.41825.65125.64125.91925.67026.64025.87725.82226.640
HVSm25.75226.67324.20226.73526.75526.70026.99326.49226.63926.62726.91026.65927.66626.87626.78627.666
Img8PSNR26.29330.41326.82430.44930.32129.96729.95028.16929.88029.82729.99629.69524.83930.11030.30530.449
Cielab3.3642.8833.3892.8612.9832.9512.9723.0162.9432.9662.9583.0205.5552.8442.8782.844
SSIM0.4680.5710.4700.5710.5680.5700.5620.5290.5700.5680.5630.5530.5820.5620.5680.582
HVS23.27526.61922.83326.88926.99226.76826.39624.37026.44626.54026.31326.22518.78826.72826.98726.992
HVSm25.22128.50524.04328.80728.95328.87728.18126.08028.52028.67428.09428.24719.10528.84028.99928.999
Img9PSNR27.26028.74228.45428.75828.68628.48928.42128.22028.68628.40928.41528.57624.59028.64828.73028.758
Cielab2.8722.6622.7442.6362.6882.8942.6992.7922.6832.9922.7022.6744.7802.6392.6492.636
SSIM0.2740.3270.2840.3270.3240.3210.3260.3130.3210.3160.3260.3180.3630.3230.3250.363
HVS23.88924.84924.60524.85724.89624.93724.49524.26324.90024.93524.48624.72619.01924.80924.91524.937
HVSm24.97525.54925.54425.56525.57525.65125.15424.99825.62125.65525.14525.49619.20325.50725.60825.655
Img10PSNR25.28527.33326.46527.32527.26827.28426.87926.67627.26427.27426.91026.91722.73127.22627.32127.333
Cielab3.9873.6113.5073.6053.6643.6173.7053.5253.6963.6243.6683.6876.1653.4723.5623.472
SSIM0.3940.4890.4120.4890.4860.4900.4840.4610.4890.4890.4830.4710.4930.4820.4890.493
HVS22.44724.03822.94523.89724.12924.22323.35323.16324.06824.18323.40223.59718.77224.02924.12824.223
HVSm24.03425.25124.38725.17125.30325.46124.53924.41625.34825.43224.59624.98019.10225.26125.35425.461
Img11PSNR25.77826.94927.42926.95026.88226.93526.70426.35726.95626.93726.69926.70523.54226.81426.93627.429
Cielab4.2434.0454.0244.0344.1084.0774.0454.2964.0574.0784.0444.0776.4634.0564.0384.024
SSIM0.4240.5130.4010.5140.5090.5190.5090.4790.5210.5190.5080.4910.5290.5030.5150.529
HVS22.25022.92623.43422.93922.97023.03222.65722.23623.01423.02222.64722.80717.56522.84822.98323.434
HVSm23.20423.69824.70823.70423.72123.79623.38723.01323.78923.79123.37523.62417.78323.61023.73824.708
Img12PSNR22.07423.21024.86423.20523.15623.18522.93722.84323.20723.21022.93723.04722.08223.16123.18624.864
Cielab5.4215.1924.0935.1915.2275.1895.1074.9535.2255.1895.1085.2097.2835.0175.1514.093
SSIM0.5120.6320.5400.6320.6290.6310.6230.5970.6320.6310.6230.6160.6610.6220.6290.661
HVS19.14119.72122.03219.71819.74519.79719.38719.43919.77919.80719.39019.65516.51919.76419.75122.032
HVSm19.93220.29823.28620.29220.30520.37819.93920.03720.36520.38919.94220.26516.73320.33120.31923.286
AveragePSNR26.20928.16827.13228.17228.08228.01327.82027.27628.03827.98627.85427.83124.54828.02328.13428.172
Cielab4.0143.6993.6663.6893.7603.7423.7403.6773.7463.7583.7333.7536.0843.5843.6623.584
SSIM0.4410.5390.4520.5390.5350.5370.5340.5050.5370.5360.5340.5230.5340.5300.5370.539
HVS22.96024.30023.18924.32824.36624.33823.91823.25924.24824.28923.95724.11419.27724.25524.38224.382
HVSm24.35725.43024.37825.46225.48825.49824.97424.33125.41525.45825.02625.32219.64425.39825.51625.516

References

  1. Bayer, B.E. Color Imaging Array. U.S. Patent 3,971,065, 20 July 1976. [Google Scholar]
  2. Bell, J.F., III; Godber, A.; McNair, S.; Caplinger, M.A.; Maki, J.N.; Lemmon, M.T.; Van Beek, J.; Malin, M.C.; Wellington, D.; Kinch, K.M.; et al. The Mars Science Laboratory Curiosity Rover Mast Camera (Mastcam) Instruments: Pre-Flight and In-Flight Calibration, Validation, and Data Archiving. AGU J. Earth Space Sci. 2017. [Google Scholar] [CrossRef]
  3. Dao, M.; Kwan, C.; Ayhan, B.; Bell, J.F. Enhancing Mastcam Images for Mars Rover Mission. In Proceedings of the 14th International Symposium on Neural Networks, Hokkaido, Japan, 21–23 June 2017; pp. 197–206. [Google Scholar]
  4. Kwan, C.; Budavari, B.; Dao, M.; Ayhan, B.; Bell, J.F. Pansharpening of Mastcam images. In Proceedings of the IEEE International Geoscience and Remote Sensing Symposium, Fort Worth, TX, USA, 23–28 July 2017; pp. 5117–5120. [Google Scholar]
  5. Ayhan, B.; Dao, M.; Kwan, C.; Chen, H.; Bell, J.F.; Kidd, R. A Novel Utilization of Image Registration Techniques to Process Mastcam Images in Mars Rover with Applications to Image Fusion, Pixel Clustering, and Anomaly Detection. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2017, 10, 4553–4564. [Google Scholar] [CrossRef]
  6. Dirik, A.E.; Memon, N. Image tamper detection based on demosaicing artifacts. In Proceedings of the 2009 16th IEEE International Conference on Image Processing (ICIP), Cairo, Egypt, 7–10 November 2009; pp. 1497–1500. [Google Scholar]
  7. Hamilton, J.; Compton, J. Processing Color and Panchromatic Pixels. U.S. Patent 20070024879A1, 1 February 2007. [Google Scholar]
  8. Kijima, T.; Nakamura, H.; Compton, J.T.; Hamilton, J.F.; DeWeese, T.E. Image Sensor with Improved Light Sensitivity. U.S. Patent 7688368B2, 30 March 2010. [Google Scholar]
  9. Zhang, C.; Li, Y.; Wang, J.; Hao, P. Universal demosaicking of color filter arrays. IEEE Trans. Image Process. 2016, 25, 5173–5186. [Google Scholar] [CrossRef] [PubMed]
  10. Condat, L. A generic variational approach for demosaicking from an arbitrary color filter array. In Proceedings of the IEEE Int. Conf. Image Process. (ICIP), Cairo, Egypt, 7–10 November 2009; pp. 1625–1628. [Google Scholar]
  11. Dokor, T.E.; King, J.; Hauptman, R. Method and Apparatus for Creating an Adaptive Bayer Pattern. U.S. Patent Application No 15/588,656, 5 March 2019. [Google Scholar]
  12. Amba, P.; Alleysson, D. LMMSE Demosaicing for multicolor CFAs. In Color and Imaging Conference; Society for Imaging Science and Technology: Paris, France, 2018; pp. 151–156. [Google Scholar]
  13. Kang, M.; Jung, M. Low-dimensional manifold model for demosaicking from a RGBW color filter array. Signal Image and Video Process. 2019. [Google Scholar] [CrossRef]
  14. Wronski, B.; Garcia-Dorado, I.; Ernst, M.; Kelly, D.; Krainin, M.; Liang, C.-K.; Levoy, M.; Milanfar, P. Handheld Multi-Frame Super-Resolution. arXiv 2019, arXiv:1905.03277. [Google Scholar] [CrossRef]
  15. Menon, D.; Calvagno, G. Regularization approaches to demosaicking. IEEE Trans. Image Process. 2009, 18, 2209–2220. [Google Scholar] [CrossRef]
  16. Kwan, C.; Chou, B.; Kwan, L.M.; Budavari, B. Debayering RGBW Color Filter Arrays: A Pansharpening Approach. In Proceedings of the IEEE Ubiquitous Computing, Electronics & Mobile Communication Conference, New York, NY, USA, 19–21 October 2017; pp. 94–100. [Google Scholar]
  17. Leung, B.; Jeon, G.; Dubois, E. Least-squares luma–chroma demultiplexing algorithm for Bayer demosaicking. IEEE Trans. Image Process. 2011, 20, 1885–1894. [Google Scholar] [CrossRef]
  18. Poisson Noise Generation. Available online: http://www.numerical-tours.com/matlab/denoisingwav_5_data_dependent/ (accessed on 22 October 2019).
  19. Rachmadi, M.F.; Valdés-Hernández, M.D.C.; Agan, M.L.F.; Di Perri, C.; Komura, T.; Alzheimer’s Disease Neuroimaging Initiative. Segmentation of white matter hyperintensities using convolutional neural networks with global spatial information in routine clinical brain MRI with none or mild vascular pathology. Comput. Med Imaging Graph. 2018, 66, 28–43. [Google Scholar] [CrossRef]
  20. Haarnoja, T.; Zhou, A.; Abbeel, P.; Levine, S. Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor. arXiv 2018, arXiv:1801.01290. [Google Scholar]
  21. Krizhevsky, A.; Sutskever, I.; Hinton, G.E. Imagenet classification with deep convolutional neural networks. In Proceedings of the Advances in Neural Information Processing Systems, Lake Tahoe, NV, USA, 3–8 December 2012; pp. 1097–1105. [Google Scholar]
  22. Saito, S.; Wei, L.; Hu, L.; Nagano, K.; Li, H. Photorealistic facial texture inference using deep neural networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), New York, NY, USA, 19–21 October 2017; Volume 3. [Google Scholar]
  23. Zhang, L.; Wu, X.; Buades, A.; Li, X. Color demosaicking by local directional interpolation and nonlocal adaptive thresholding. J. Electron. Imaging 2011, 20, 023016. [Google Scholar]
  24. Malvar, H.S.; He, L.-W.; Cutler, R. High-quality linear interpolation for demosaciking of color images. In Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, Montreal, QC, Canada, 17–21 May 2004; pp. 485–488. [Google Scholar]
  25. Zhang, L.; Wu, X. Color demosaicking via directional linear minimum mean square-error estimation. IEEE Trans. Image Process. 2005, 14, 2167–2178. [Google Scholar] [CrossRef]
  26. Lu, W.; Tan, Y.P. Color filter array demosaicking: New method and performance measures. IEEE Trans. Image Process. 2003, 12, 1194–1210. [Google Scholar] [PubMed]
  27. Dubois, E. Frequency-domain methods for demosaicking of Bayer-sampled color images. IEEE Signal Proc. Letters 2005, 12, 847–850. [Google Scholar] [CrossRef]
  28. Gunturk, B.; Altunbasak, Y.; Mersereau, R.M. Color plane interpolation using alternating projections. IEEE Transactions Image Process. 2002, 11, 997–1013. [Google Scholar] [CrossRef] [PubMed]
  29. Wu, X.; Zhang, N. Primary-consistent soft-decision color demosaicking for digital cameras. IEEE Trans. Image Process. 2004, 13, 1263–1274. [Google Scholar] [CrossRef]
  30. Kwan, C.; Chou, B.; Kwan, L.M.; Larkin, J.; Ayhan, B.; Bell, J.F.; Kerner, H. Demosaicking enhancement using pixel-level fusion. J. Signal Image Video Process. 2018, 12, 749–756. [Google Scholar] [CrossRef]
  31. Bednar, J.; Watt, T. Alpha-trimmed means and their relationship to median filters. IEEE Trans. Acoust. Speech Signal Process. 1984, 32, 145–153. [Google Scholar] [CrossRef]
  32. Gharbi, M.; Chaurasia, G.; Paris, S.; Durand, F. Deep joint demosaicking and denoising. ACM Trans. Graph 2016, 35, 191. [Google Scholar] [CrossRef]
  33. Klatzer, T.; Hammernik, K.; Knobelreiter, P.; Pock, T. Learning joint demosaicing and denoising based on sequential energy minimization. In Proceedings of the IEEE International Conference on Computational Photography (ICCP), Evanston, IL, USA, 13–15 May 2016; pp. 1–11. [Google Scholar]
  34. Jaiswal, S.P.; Au, O.C.; Jakhetiya, V.; Yuan, Y.; Yang, H. Exploitation of inter-color correlation for color image demosaicking. In Proceedings of the 2014 IEEE International Conference on Image Processing (ICIP), Paris, France, 27–30 October 2014; pp. 1812–1816. [Google Scholar]
  35. Kiku, D.; Monno, Y.; Tanaka, M.; Okutomi, M. Beyond color difference: Residual interpolation for color image demosaicking. IEEE Trans. Image Process. 2016, 25, 1288–1300. [Google Scholar] [CrossRef]
  36. Monno, Y.; Kiku, D.; Tanaka, M.; Okutomi, M. Adaptive Residual Interpolation for Color and Multispectral Image Demosaicking. Sensors 2017, 17, 2787. [Google Scholar] [CrossRef]
  37. Wu, J.; Timofte, R.; Gool, L.V. Demosaicing based on directional difference regression and efficient regression priors. IEEE Trans. Image Process. 2016, 25, 3862–3874. [Google Scholar] [CrossRef]
  38. Losson, O.; Macaire, L.; Yang, Y. Comparison of Color Demosaicing Methods. In Advances in Imaging and Electron Physics; Peter, W.H., Ed.; Elsevier: Amsterdam, The Netherlands, 2010; Volume 162, pp. 173–265. [Google Scholar]
  39. Vivone, G.; Alparone, L.; Chanussot, J.; Dalla Mura, M.; Garzelli, A.; Licciardi, G.A.; Restaino, R.; Wald, L. A Critical Comparison Among Pansharpening Algorithms. IEEE Trans. Geosci. Remote Sens. 2015, 53, 2565–2586. [Google Scholar] [CrossRef]
  40. Liu, J.G. Smoothing filter based intensity modulation: A spectral preserve image fusion technique for improving spatial details. Int. J. Remote Sens. 2000, 21, 18. [Google Scholar] [CrossRef]
  41. Aiazzi, B.; Alparone, L.; Baronti, S.; Garzelli, A.; Selva, M. MTF-tailored multiscale fusion of high-resolution MS and pan imagery. Photogramm. Eng. Remote Sens. 2006, 72, 591–596. [Google Scholar]
  42. Vivone, G.; Restaino, R.; Dalla Mura, M.; Licciardi, G.; Chanussot, J. Contrast and error-based fusion schemes for multispectral image pansharpening. IEEE Trans. Geosci. Remote Sens. Lett. 2014, 11, 930–934. [Google Scholar] [CrossRef]
  43. Laben, C.; Brower, B. Process for Enhancing the Spatial Resolution of Multispectral Imagery Using Pan-Sharpening. U.S. Patent 6 011 875, 4 January 2000. [Google Scholar]
  44. Aiazzi, B.; Baronti, S.; Selva, M. Improving component substitution pansharpening through multivariate regression of MS+pan data. IEEE Trans. Geosci. Remote Sens. 2007, 45, 3230–3239. [Google Scholar] [CrossRef]
  45. Liao, W.; Huang, X.; Van Coillie, F.; Gautama, S.; Pižurica, A.; Philips, W.; Hui, L.; Tingting, Z.; Shimoni, M.; Tuia, D.; et al. Processing of multiresolution thermal hyperspectral and digital color data: Outcome of the 2014 IEEE GRSS data fusion contest. IEEE J. Select. Top. Appl. Earth Observ. Remote Sens. 2015, 8, 6. [Google Scholar] [CrossRef]
  46. Choi, J.; Yu, K.; Kim, Y. A new adaptive component-substitution based satellite image fusion by using partial replacement. IEEE Trans. Geosci. Remote Sens. 2011, 49, 295–309. [Google Scholar] [CrossRef]
  47. Zhou, J.; Kwan, C.; Budavari, B. Hyperspectral image super-resolution: A hybrid color mapping approach. J. Appl. Remote Sens. 2016, 10, 035024. [Google Scholar] [CrossRef]
  48. Kwan, C.; Choi, J.H.; Chan, S.; Zhou, J.; Budavai, B. Resolution Enhancement for Hyperspectral Images: A Super-Resolution and Fusion Approach. In Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, New Orleans, LA, USA, 5–9 March 2017; pp. 6180–6184. [Google Scholar]
  49. Kwan, C.; Budavari, B.; Feng, G. A Hybrid Color Mapping Approach to Fusing MODIS and Landsat Images for Forward Prediction. Remote Sens. 2017, 10, 520. [Google Scholar] [CrossRef]
  50. Kwan, C.; Budavari, B.; Bovik, A.; Marchisio, G. Blind Quality Assessment of Fused WorldView-3 Images by Using the Combinations of Pansharpening and Hypersharpening Paradigms. IEEE Geosci. Remote Sens. Lett. 2017, 14, 1835–1839. [Google Scholar] [CrossRef]
  51. Kwan, C.; Ayhan, B.; Budavari, B. Fusion of THEMIS and TES for Accurate Mars Surface Characterization. In Proceedings of the IEEE International Geoscience and Remote Sensing Symposium, Fort Worth, TX, USA, 23–28 July 2017; pp. 3381–3384. [Google Scholar]
  52. Kwan, C.; Chou, B. Further improvement of debayering performance of RGBW color filter arrays using deep learning and pansharpening techniques. J. Imaging 2019, 5, 68. [Google Scholar] [CrossRef]
  53. Rafinazaria, M.; Dubois, E. Demosaicking algorithm for the Kodak-RGBW color filter array. In Color Imaging XX: Displaying, Processing, Hardcopy, and Applications, Proceedings of the SPIE-IS&T Electronic Imaging, San Francisco, CA, USA, 8–12 February 2015; International Society for Optics and Photonics: Bellingham, WA, USA; Volume 9395, p. 929503.
  54. BM3D Denoising. Available online: http://www.cs.tut.fi/~foi/invansc/ (accessed on 22 October 2019).
  55. Kwan, C.; Zhu, X.; Gao, F.; Chou, B.; Perez, D.; Li, J.; Shen, Y.; Koperski, K. Assessment of Spatiotemporal Fusion Algorithms for Planet and Worldview Images. Sensors 2018, 18, 1051. [Google Scholar] [CrossRef]
  56. SSIM. Available online: https://en.wikipedia.org/wiki/Structural_similarity (Accessed on 26 April 2019).
  57. Ochoa, H.D.; Rao, K.R. Discrete Cosine Transform, II ed.; CRC Press: Boca Raton, FL, USA, 2019. [Google Scholar]
  58. Egiazarian, K.; Astola, J.; Ponomarenko, N.; Lukin, V.; Battisti, F.; Carli, M. New full quality metrics based on HVS. In Proceedings of the Second International Workshop on Video Processing and Quality Metrics, Scottsdale, AZ, USA, 22–24 January 2006. [Google Scholar]
  59. Ponomarenko, N.; Silvestri, F.; Egiazarian, K.; Carli, M.; Astola, J.; Lukin, V. On between-coefficient contrast masking of DCT basis functions. In Proceedings of the Third International Workshop on Video Processing and Quality Metrics for Consumer Electronics VPQM-07, Scottsdale, AZ, USA, 25–26 January 2007. [Google Scholar]
  60. Damera-Venkata, N.; Kite, T.D.; Geisler, W.S.; Evans, B.L.; Bovik, A.C. Image quality assessment based on a degradation model. IEEE Trans. Image Process. 2000, 9, 636–650. [Google Scholar] [CrossRef]
  61. Watson, A.B.; Solomon, J.A.; Ahumada, A.J.; Gale, A. Discrete cosine transform (DCT) basis function visibility: Effects of viewing distance and contrast masking. In Proceedings of the SPIE 2179, Human Vision, Visual Processing, and Digital Display V, San Jose, CA, USA, 8–10 February 1994. [Google Scholar]
  62. Wang, Z.; Bovik, A.C. A universal image quality index. IEEE Signal Process. Lett. 2002, 9, 81–84. [Google Scholar] [CrossRef]
  63. Sheikh, H.R.; Bovik, A.C. Image information and visual quality. IEEE Trans. Image Process. 2006, 15, 430–444. [Google Scholar] [CrossRef]
  64. Miyahara, M.; Kotani, K.; Algazi, V.R. Objective picture quality scale (PQS) for image coding. IEEE Trans. Commun. 1998, 46, 1215–1226. [Google Scholar] [CrossRef]
  65. Zhang, X.; Wandell, B.A. A spatial extension of cielab for digital color image reproduction. J. Soc. Inf. Disp. 1997, 5, 61–63. [Google Scholar] [CrossRef]
  66. Kwan, C. Active Noise Reduction System for Creating a Quiet Zone. U.S. Patent #9773494, 27 September 2017. [Google Scholar]
  67. Kwan, C.; Zhou, J.; Qiao, J.; Liu, G.; Ayhan, B. A High Performance Approach to Local Active Noise Reduction. In Proceedings of the IEEE Conference on Decision and Control, Las Vegas, NV, USA, 12–14 December 2016; pp. 347–352. [Google Scholar]
  68. Kwan, C.; Chu, S.; Yin, J.; Liu, X.; Kruger, M.; Sityar, I. Enhanced speech in noisy multiple speaker environment. In Proceedings of the IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence), Hong Kong, China, 1–8 June 2008. [Google Scholar]
Figure 1. Two color filter array (CFA) patterns. (a) Standard Bayer pattern (CFA 1.0); (b) RGBW (red-green-blue-white) (CFA 2.0).
Figure 1. Two color filter array (CFA) patterns. (a) Standard Bayer pattern (CFA 1.0); (b) RGBW (red-green-blue-white) (CFA 2.0).
Electronics 08 01444 g001
Figure 2. Standard approach to demosaicing CFA 2.0 images. Image from [38].
Figure 2. Standard approach to demosaicing CFA 2.0 images. Image from [38].
Electronics 08 01444 g002
Figure 3. A pansharpening approach to demosaicing CFA 2.0 images [16].
Figure 3. A pansharpening approach to demosaicing CFA 2.0 images [16].
Electronics 08 01444 g003
Figure 4. A combination of deep learning and pansharpening approach to demosaicing CFA 2.0.
Figure 4. A combination of deep learning and pansharpening approach to demosaicing CFA 2.0.
Electronics 08 01444 g004
Figure 5. Twelve clean images from the Kodak dataset.
Figure 5. Twelve clean images from the Kodak dataset.
Electronics 08 01444 g005aElectronics 08 01444 g005b
Figure 6. Twelve noisy images at 10 dB (Poisson noise).
Figure 6. Twelve noisy images at 10 dB (Poisson noise).
Electronics 08 01444 g006aElectronics 08 01444 g006b
Figure 7. Twelve noisy images at 20 dB (Poisson noise).
Figure 7. Twelve noisy images at 20 dB (Poisson noise).
Electronics 08 01444 g007
Figure 8. Noisy images using Gaussian additive noise at a signal-to-noise ratio (SNR) of 10 dBs. The image characteristics are very different from those of Poisson noise.
Figure 8. Noisy images using Gaussian additive noise at a signal-to-noise ratio (SNR) of 10 dBs. The image characteristics are very different from those of Poisson noise.
Electronics 08 01444 g008
Figure 9. Summary of all experiments and their corresponding sections.
Figure 9. Summary of all experiments and their corresponding sections.
Electronics 08 01444 g009
Figure 10. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 1.0 pattern. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 10. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 1.0 pattern. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g010aElectronics 08 01444 g010b
Figure 11. Visual comparison of three high performing demosaicing algorithms for CFA 1.0 at 10 dBs SNR (Poisson noise).
Figure 11. Visual comparison of three high performing demosaicing algorithms for CFA 1.0 at 10 dBs SNR (Poisson noise).
Electronics 08 01444 g011
Figure 12. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 1.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 12. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 1.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g012
Figure 13. Visual comparison of three demosaicing results for CFA 1.0 at 10 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Figure 13. Visual comparison of three demosaicing results for CFA 1.0 at 10 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Electronics 08 01444 g013
Figure 14. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 1.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 14. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 1.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g014
Figure 15. Visual comparison of three demosaicing results for CFA 1.0 at 10 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Figure 15. Visual comparison of three demosaicing results for CFA 1.0 at 10 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Electronics 08 01444 g015
Figure 16. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 1.0 pattern without denoising. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 16. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 1.0 pattern without denoising. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g016aElectronics 08 01444 g016b
Figure 17. Visual comparison of three high performing demosaicing algorithms for CFA 1.0 at 20 dBs SNR (Poisson noise). No denoising.
Figure 17. Visual comparison of three high performing demosaicing algorithms for CFA 1.0 at 20 dBs SNR (Poisson noise). No denoising.
Electronics 08 01444 g017
Figure 18. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 1.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 18. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 1.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g018
Figure 19. Visual comparison of three demosaicing results for CFA 1.0 at 20 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Figure 19. Visual comparison of three demosaicing results for CFA 1.0 at 20 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Electronics 08 01444 g019
Figure 20. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 1.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 20. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 1.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g020
Figure 21. Visual comparison of three demosaicing results for CFA 1.0 at 20 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Figure 21. Visual comparison of three demosaicing results for CFA 1.0 at 20 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Electronics 08 01444 g021
Figure 22. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 2.0 pattern without denoising. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 22. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 2.0 pattern without denoising. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g022
Figure 23. Visual comparison of three high performing demosaicing algorithms for CFA 2.0 at 10 dBs SNR (Poisson noise). No denoising.
Figure 23. Visual comparison of three high performing demosaicing algorithms for CFA 2.0 at 10 dBs SNR (Poisson noise). No denoising.
Electronics 08 01444 g023aElectronics 08 01444 g023b
Figure 24. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 2.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 24. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 2.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g024
Figure 25. Visual comparison of three demosaicing results for CFA 2.0 at 10 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Figure 25. Visual comparison of three demosaicing results for CFA 2.0 at 10 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Electronics 08 01444 g025
Figure 26. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 2.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 26. Averaged performance metrics for all the low light images at 10 dBs SNR (Poisson noise) using CFA 2.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g026
Figure 27. Visual comparison of three demosaicing results for CFA 2.0 at 10 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Figure 27. Visual comparison of three demosaicing results for CFA 2.0 at 10 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Electronics 08 01444 g027
Figure 28. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 2.0 pattern without denoising. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 28. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 2.0 pattern without denoising. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g028aElectronics 08 01444 g028b
Figure 29. Visual comparison of three high performing demosaicing algorithms for CFA 2.0 at 20 dBs SNR (Poisson noise). No denoising.
Figure 29. Visual comparison of three high performing demosaicing algorithms for CFA 2.0 at 20 dBs SNR (Poisson noise). No denoising.
Electronics 08 01444 g029
Figure 30. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 2.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 30. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 2.0 pattern with denoising after demosaicing. (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g030
Figure 31. Visual comparison of three demosaicing results for CFA 2.0 at 20 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Figure 31. Visual comparison of three demosaicing results for CFA 2.0 at 20 dBs SNR (Poisson noise). Denoising is performed after demosaicing.
Electronics 08 01444 g031
Figure 32. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 2.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Figure 32. Averaged performance metrics for all the low light images at 20 dBs SNR (Poisson noise) using CFA 2.0 pattern. Denoising is after CFA is generated and before demosaicing starts: (a) PNSR; (b) CIELAB; (c) SSIM; (d) HVS and HVSm.
Electronics 08 01444 g032
Figure 33. Visual comparison of three demosaicing results for CFA 2.0 at 20 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Figure 33. Visual comparison of three demosaicing results for CFA 2.0 at 20 dBs SNR (Poisson noise). Denoising is performed after CFA is generated and before demosaicing starts.
Electronics 08 01444 g033
Figure 34. Best against the best comparison between CFA 1.0 and CFA 2.0 with and without denoising at 10 dBs SNR. (a) PSNR; (b) Cielab; (c) SSIM; (d) HVS; (e) HVSm.
Figure 34. Best against the best comparison between CFA 1.0 and CFA 2.0 with and without denoising at 10 dBs SNR. (a) PSNR; (b) Cielab; (c) SSIM; (d) HVS; (e) HVSm.
Electronics 08 01444 g034
Figure 35. Best against the best comparison between CFA 1.0 and CFA 2.0 with and without denoising at 20-dB SNR. (a) PSNR; (b) Cielab; (c) SSIM; (d) HVS; (e) HVSm.
Figure 35. Best against the best comparison between CFA 1.0 and CFA 2.0 with and without denoising at 20-dB SNR. (a) PSNR; (b) Cielab; (c) SSIM; (d) HVS; (e) HVSm.
Electronics 08 01444 g035
Figure 36. No denoising case at 10 dBs. Error distributions of the two CFAs.
Figure 36. No denoising case at 10 dBs. Error distributions of the two CFAs.
Electronics 08 01444 g036
Figure 37. A hypothetical optimization problem.
Figure 37. A hypothetical optimization problem.
Electronics 08 01444 g037
Table 1. Correlation of different metrics to human’s visual perception.
Table 1. Correlation of different metrics to human’s visual perception.
MeasureReferenceSpearman CorrelationKendall Correlation
HVS-mPonomarenko, N.; et al. [59]0.9840.948
HVSEgiazarian, K.; et al. [58]0.8950.712
NQMDamera-Venkata, N.; et al. [60]0.8570.673
DCTuneWatson, A. B.; et al. [61] 0.8290.712
UQIWang, Z.; Bovik, A. C. [62]0.5500.438
PSNRPeak Signal to Noise Ratio [55]0.5370.359
SSIMStructural similarity [56]0.4060.358
VIFSheikh, H. R.; Bovik, A. C. [63]0.3770.255
PQSMiyahara, M.; Kotani, K.; Algazi, V. R. [64]0.3020.242
Table 2. Comparison of CFA patterns for the various demosaicing cases at 10 dBs SNR. Bold numbers indicate the high performance CFA pattern.
Table 2. Comparison of CFA patterns for the various demosaicing cases at 10 dBs SNR. Bold numbers indicate the high performance CFA pattern.
MetricsCFANo Denoising/Best AlgorithmDenoising After Demosaicing/Best AlgorithmDenoising Before Demosaicing/Best Algorithm
PSNR (dB)1.016.889/F320.826/F321.978/F3
2.021.249/F324.050/LSLCD26.141/Demonet+GFPCA
CIELAB1.010.149/GFPCA6.664/F36.545/Demonet
2.06.354/GFPCA5.516/F34.310/Demonet+GFPCA
SSIM1.00.455/F30.476/ATMF0.463/ATMF
2.00.451/ATMF0.459/LSLCD0.467/Standard
HVS (dB)1.012.285/SEM16.229/F316.833/ARI
2.016.531/F319.056/LSLCD22.053/Demonet+GFPCA
HVSm (dB)1.012.403/SEM16.494/F317.116/ARI
2.016.868/F319.568/LSLCD23.121/Demonet+GFPCA
Table 3. Comparison of CFA patterns for the various demosaicing cases at 20 dBs SNR. Bold numbers indicate the high performance CFA pattern.
Table 3. Comparison of CFA patterns for the various demosaicing cases at 20 dBs SNR. Bold numbers indicate the high performance CFA pattern.
MetricsCFANo Denoising/Best AlgorithmDenoising After Demosaicing/Best AlgorithmDenoising Before Demosaicing/Best Algorithm
PSNR (dB)1.020.488/ATMF22.821/F324.059/Bilinear
2.023.290/F324.391/GSA28.172/LSLCD
CIELAB1.06.713/Demonet5.256/Demonet4.935/Demonet
2.05.121/GFPCA5.268/LSLCD3.584/F3
SSIM1.00.500/SEM0.548/F30.574/F3
2.00.545/Demonet+GFPCA0.535/LSLCD0.539/GSA
HVS (dB)1.016.130/Demonet18.204/Bilinear19.142/Demonet
2.018.646/F319.415/LSLCD24.382/ATMF
HVSm (dB)1.016.365/Demonet18.734/Bilinear19.444/ARI
2.019.112/F319.881/LSLCD25.516/ATMF

Share and Cite

MDPI and ACS Style

Kwan, C.; Larkin, J. Demosaicing of Bayer and CFA 2.0 Patterns for Low Lighting Images. Electronics 2019, 8, 1444. https://doi.org/10.3390/electronics8121444

AMA Style

Kwan C, Larkin J. Demosaicing of Bayer and CFA 2.0 Patterns for Low Lighting Images. Electronics. 2019; 8(12):1444. https://doi.org/10.3390/electronics8121444

Chicago/Turabian Style

Kwan, Chiman, and Jude Larkin. 2019. "Demosaicing of Bayer and CFA 2.0 Patterns for Low Lighting Images" Electronics 8, no. 12: 1444. https://doi.org/10.3390/electronics8121444

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop