Search Results (7)

Recent advanced systems in the speech recognition domain use large Transformer neural networks that have been pretrained on massive speech data. General methods in the deep learning area have been frequently shared across various domains, and the Transformer model can also be used effectively across speech and image. In this paper, we introduce a novel masking method for self-supervised speech representation learning with salt-and-pepper (S&P) mask which is commonly used in computer vision. The proposed scheme includes consecutive quadrilateral-shaped S&P patches randomly contaminating the input speech spectrum. Furthermore, we modify the standard S&P mask to make it appropriate for the speech domain. In order to validate the effect of the proposed spectral S&P patch masking for the self-supervised representation learning approach, we conduct the pretraining and downstream experiments with two languages, English and Korean. To this end, we pretrain the speech representation model using each dataset and evaluate the pretrained models for feature extraction and fine-tuning performance on varying downstream tasks, respectively. The experimental outcomes clearly illustrate that the proposed spectral S&P patch masking is effective for various downstream tasks when combined with the conventional masking methods. Full article

Salt-and-pepper noise (SPN) is a common type of image noise that appears as randomly distributed white and black pixels in an image. It is also known as impulse noise or random noise. This paper aims to introduce a new weighted average based on the Atangana–Baleanu fractional integral operator, which is a well-known idea in fractional calculus. Our proposed method also incorporates the concept of symmetry in the window mask structures, resulting in efficient and easily implementable filters for real-time applications. The distinguishing point of these techniques compared to similar methods is that we employ a novel idea for calculating the mean of regular pixels rather than the existing used mean formula along with the median. An iterative procedure has also been provided to integrate the power of removing high-density noise. Moreover, we will explore the different approaches to image denoising and their effectiveness in removing noise from images. The symmetrical structure of this tool will help in the ease and efficiency of these techniques. The outputs are compared in terms of peak signal-to-noise ratio, the mean-square error and structural similarity values. It was found that our proposed methodologies outperform some well-known compared methods. Moreover, they boast several advantages over alternative denoising techniques, including computational efficiency, the ability to eliminate noise while preserving image features, and real-time applicability. Full article

With the advent of unsupervised learning, efficient training of a deep network for image denoising without pairs of noisy and clean images has become feasible. Most current unsupervised denoising methods are built on self-supervised loss with the assumption of zero-mean noise under the signal-independent condition, which causes brightness-shifting artifacts on unconventional noise statistics (i.e., different from commonly used noise models). Moreover, most blind denoising methods require a random masking scheme for training to ensure the invariance of the denoising process. In this study, we propose a dilated convolutional network that satisfies an invariant property, allowing efficient kernel-based training without random masking. We also propose an adaptive self-supervision loss to increase the tolerance for unconventional noise, which is specifically effective in removing salt-and-pepper or hybrid noise where prior knowledge of noise statistics is not readily available. We demonstrate the efficacy of the proposed method by comparing it with state-of-the-art denoising methods using various examples. Full article

Accurate cropland burned area estimation is crucial for air quality modeling and cropland management. However, current global burned area products have been primarily derived from coarse spatial resolution images which cannot fulfill the spatial requirement for fire monitoring at local levels. In addition, there is an overall lack of accurate cropland straw burning identification approaches at high temporal and spatial resolution. In this study, we propose a novel algorithm to capture burned area in croplands using dense Landsat time series image stacks. Cropland burning shows a short-term seasonal variation and a long-term dynamic trend, so a multi-harmonic model is applied to characterize fire dynamics in cropland areas. By assessing a time series of the Burned Area Index (BAI), our algorithm detects all potential burned areas in croplands. A land cover mask is used on the primary burned area map to remove false detections, and the spatial information with a moving window based on a majority vote is employed to further reduce salt-and-pepper noise and improve the mapping accuracy. Compared with the accuracy of 67.3% of MODIS products and that of 68.5% of Global Annual Burned Area Map (GABAM) products, a superior overall accuracy of 92.9% was obtained by our algorithm using Landsat time series and multi-harmonic model. Our approach represents a flexible and robust way of detecting straw burning in complex agriculture landscapes. In future studies, the effectiveness of combining different spectral indices and satellite images can be further investigated. Full article

Edge detection is one of the main steps in the image processing field, especially in biomedical imaging, to diagnose a disease or trace its progress. The transfer of medical images makes them more susceptible to quality degradation due to any imposed noise. Hence, the protection of this data against noise is a persistent need. The efficiency of fractional-order filters to detect fine details and their high noise robustness, unlike the integer-order filters, it renders them an attractive solution for biomedical edge detection. In this work, two novel central fractional-order masks are proposed with their detailed mathematical proofs. The fractional-order parameter gives an extra degree of freedom in designing different masks. The noise performance of the proposed masks is evaluated upon applying Salt and Pepper noise and Gaussian noise. Numerical results proved that the proposed masks outperform the integer-order masks regarding both types of noise, achieving higher Peak Signal to Noise Ratio. As a practical application, the proposed fractional-order edge detection masks are employed to enhance the Diabetic Retinopathy disease diagnosis. Full article

Conventional quantization-based watermarking may be easily estimated by averaging on a set of watermarked signals via uniform quantization approach. Moreover, the conventional quantization-based method neglects the visual perceptual characteristics of the host signal; thus, the perceptible distortions would be introduced in some parts of host signal. In this paper, inspired by the Watson’s entropy masking model and logarithmic quantization index modulation (LQIM), a logarithmic quantization-based image watermarking method is developed by using the wavelet transform. Furthermore, the novel method improves the robustness of watermarking based on a logarithmic quantization strategy, which embeds the watermark data into the image blocks with high entropy value. The main significance of this work is that the trade-off between invisibility and robustness is simply addressed by using the logarithmic quantizaiton approach, which applies the entropy masking model and distortion-compensated scheme to develop a watermark embedding method. In this manner, the optimal quantization parameter obtained by minimizing the quantization distortion function effectively controls the watermark strength. In terms of watermark decoding, we model the wavelet coefficients of image by the generalized Gaussian distribution (GGD) and calculate the bit error probability of proposed method. Performance of the proposed method is analyzed and verified by simulation on real images. Experimental results demonstrate that the proposed method has the advantages of imperceptibility and strong robustness against attacks covering JPEG compression, additive white Gaussian noise (AWGN), Gaussian filtering, Salt&Peppers noise, scaling and rotation attack, etc. Full article

This study aims to detect coastline changes using temporal synthetic aperture radar (SAR) images for the state of Kelantan, Malaysia. Two active images, namely, RADARSAT-1 captured in 2003 and RADARSAT-2 captured in 2014, were used to monitor such changes. We applied noise removal and edge detection filtering on RADARSAT images for preprocessing to remove salt and pepper distortion. Different segmentation analyses were also applied to the filtered images. Firstly, multiresolution segmentation, maximum spectral difference and chessboard segmentation were performed to separate land pixels from ocean ones. Next, the Taguchi method was used to optimise segmentation parameters. Subsequently, a support vector machine algorithm was applied on the optimised segments to classify shorelines with an accuracy of 98% for both temporal images. Results were validated using a thematic map from the Department of Survey and Mapping of Malaysia. The change detection showed an average difference in the shoreline of 12.5 m between 2003 and 2014. The methods developed in this study demonstrate the ability of active SAR sensors to map and detect shoreline changes, especially during low or high tides in tropical regions where passive sensor imagery is often masked by clouds. Full article