Lossless Hyperspectral Image Compression in Comet Interceptor and Hera Missions with Restricted Bandwith

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The core work of this paper is to compare the performance of CCSDS122.0 and JPEG2000 in hyperspectral compression. The article is generally well-structured and experimental. But they did not propose their own innovative work, and the work lacks innovation.

1. In the background presentation, the presentation on CCSDS122.0 and JPEG2000 was not sufficient, and needed to be supplemented with more detailed information and more high-quality related work.

2. In the 2.5 Noise Filtering, there are four types of noise filtering, three of which are wavelet-based, and the repetition rate is too high. One or two other types of filtering should be selected for relevant experiments to increase the reliability of the experiment.

3. The image in Figure 6 can be partially enlarged so that the reader can compare the differences.

4. Although the experiments in this paper are abundant, the specific analysis and data presentation are lacking. For example, specific parameters, complexity, compression time, and other information should be displayed in the form of tables.

5. Conclusions are written too briefly, and specific data should be added appropriately, such as specific performance and parameter differences, rather than vague descriptions.

Author Response

We thank you for spending your time carefully reviewing the manuscript and for your opinions regarding the science and presentation of the material.

Comment 1: In the background presentation, the presentation on CCSDS122.0 and JPEG2000 was not sufficient, and needed to be supplemented with more detailed information and more high-quality related work.

Response 1: Thank you for pointing this out. We present a broad overview of both algorithms and refer to the key references relevant to the mathematics and architecture behind the algorithms used. However, we are open to any suggestions on any more suitable ones. Implementation details are discussed separately, and we added information about the parameters of CCSDS 122.0 and the code complexity of both algorithms. 

The first updated line can be found on page 7, paragraph 2, line 205. "These changes enable the source code to only consist of 1513 total lines of code that compile to a binary of 35 KB. "

Similar information was also added for JPEG 2000 on page 7, paragraph 6, line 234. "The JasPer library contains 34413 lines of code that produce a binary of 548 KB."

We also added paragraph four on the same page. "The configuration parameter choices were mostly based on the constraints set by lossless compression. Neither a byte limit nor any early stopping flags were enabled to guarantee the encoding of all data. Notably, the heuristic methods for the bitplane encoder are used. This change did not negatively affect the compression performance of the algorithm but decreased both memory usage and speed significantly. Moreover, the subband weights, described in the CCSDS 122.0 standard, were used. No alterations to these parameters were conducted during testing or validation."  

Comment 2: In the 2.5 Noise Filtering, there are four types of noise filtering, three of which are wavelet-based, and the repetition rate is too high. One or two other types of filtering should be selected for relevant experiments to increase the reliability of the experiment.

Response 2: The focus of our manuscript is not to compare different noise filtering methods but to compare the performance of both algorithms in different noise scenarios. The filters were introduced to have an imagery dataset with variable amounts of noise (some filters remove noise better than others) to represent different 'real world' scenarios. This enables us to fill the gap between noisy (unfiltered) and ideal noiseless scenarios. We added a note about it in the manuscript to clarify this to the readers. A detailed manuscript on noise filtering of hyperspectral data is under preparation by one of the co-authors.

The changes can be found on lines 299 and 300 in the last paragraph on page 9. "The compression performance tests and benchmarking against JPEG 2000 JasPer codec followed the successful verification of the CCSDS 122.0 implementation. Each data point was calculated as the average compression ratio of all wavelengths corresponding to an image, 10 for Vis and 20 for NIR. The whole datacube was divided into two parts: ASPECT Vis and NIR1 (2 channels). Next, all wavelengths were compressed as single 2D images by both the JPEG 2000 and CCSDS 122.0 algorithms. Differentially encoded images were first processed with a script provided by the Brno University of Technology before compression. Additionally, multiple noise filters were used to reduce noise by varying amounts. These filtered images are used to provide data comparable to real-world scenarios and increase coverage of noise scenarios. Finally, noiseless versions of each image (see Figure 7) were compressed to act as a sanity check and provide a reference level for the maximum achievable compression ratio."

Comment 3: The image in Figure 6 can be partially enlarged so that the reader can compare the differences.

Response 3: We agree with this comment and have enlarged all three images accordingly. The updated Figure 7 can be found on page 10.

Comment 4: Although the experiments in this paper are abundant, the specific analysis and data presentation are lacking. For example, specific parameters, complexity, compression time, and other information should be displayed in the form of tables.

Response 4: We agree and have added tables containing compression time, memory usage, and image scenarios. The added Figure 4 can be found on page 6 and the tables 1 and 2 with corresponding paragraphs on page 14. 

Comment 5. Conclusions are written too briefly, and specific data should be added appropriately, such as specific performance and parameter differences, rather than vague descriptions.

Response 5: We agree with this comment and have extended the conclusions and provided a quantitative comparison of the two algorithms. The updated conclusion section can be found on page 16. 
"The JPEG 2000 and CCSDS 122.0 algorithms were implemented and tested on synthetic hyperspectral images of Didymos-Dimorphos asteroids. Both algorithms show a consistent reduction in hyperspectral image data volume. A few trends were observed. While CCSDS 122.0 performed on average 1-2 pp better in compressing the noise-containing images, JPEG 2000 offered 1-2 pp greater compression performance of the noise-filtered images. The original noiseless images were compressed similarly with both algorithms. Compared to JPEG 2000, CCSDS 122.0 has the advantages of an over 15 times smaller binary code size and approx. 1.3 times faster image compression. Regarding memory usage, JPEG 2000 has the advantage of dynamic bit depth reduction of its internal data structures from 16 to 8 bits when the image entropy allows this. For this reason, JPEG 2000 Vis image compression used half the memory compared to the CCSDS 122.0 case, where 16-bit was kept. With NIR images, both algorithms worked in full 16-bit mode, resulting in similar memory demands. Additional noise filtering, especially wavelet 3D-based algorithms such as FORPDN or HyRes, reduced the image size by half. The benefit of differential encoding is the reduction of asteroid coverage dependence on asteroid vs. space background, resulting in more consistent and predictable data volume savings. When choosing an image compression algorithm, the advantages and disadvantages exhibited by both algorithms need to be weighed on a case-by-case basis, and further tests should be performed on mission-specific hardware."

Reviewer 2 Report

Comments and Suggestions for Authors

Lossless image compression is very important for faster transmission. This manuscript compared performance of two image compression algorithms, CCSDS 122.0 and JPEG 2000 in varying scenarios, and provided usage recommendations. Here are my suggestions.

(1)   The paper compares two image compression methods, CCSDS 122.0 and JPEG 2000. Each compression method has specific strengths or limitations and is recommended for different scenarios. Is it possible to propose an improved method more suitable for the mission data based on current analysis?

(2)   It is recommended to supplement the specific descriptions of different scenarios and clarify the specific requirements of different scenarios for image compression.

(3)   In Section 2, the Asteroid Image Simulator is the key tool for generating simulated data. It is recommended to supplement the detailed descriptions of this tool and add a flowchart illustrating the process of simulated data generation.

(4)   Section 4, Discussion: The analysis in this section is too broad. It is recommended to conduct a more quantitative comparison of the performance of these two methods, and further explore the boundary conditions of each compression method used in different scenario.

(5)   Spectral information is extremely crucial for hyperspectral data. Could you provide a more detailed analysis about the ability of these two compression methods to preserve and reconstruct spectral information?

Author Response

We thank you for spending your time carefully reviewing the manuscript and for your opinions regarding the science and presentation of the material.

Comment 1: The paper compares two image compression methods, CCSDS 122.0 and JPEG 2000. Each compression method has specific strengths or limitations and is recommended for different scenarios. Is it possible to propose an improved method more suitable for the mission data based on current analysis?

Response 1: While recommending a totally new compression method not based on bitplane encoders is not possible based on current analysis, we added discussion and appropriate references about a preprocessing method called spectral unmixing autoencoders as a possible topic for future research. 

The added paragraph can be found in the section 4.2 "Limitations and Future Prospects" on page 15. "Alternative options for preprocessing methods should also be considered in the future. One potential method consists of spectral unmixing autoencoders (SUA). These SUAs can be modeled using neural networks performing differential encoding and noise reduction. This reduces computational complexity as neural networks do not require matrix decompositions or other advanced matrix operations. However, training these networks requires a significant amount of tailored data, which is not easily modeled or readily available." 

Comment 2: It is recommended to supplement the specific descriptions of different scenarios and clarify the specific requirements of different scenarios for image compression.

Response 2: We agree that the differences between scenarios need to be demonstrated and have added descriptions of the simulated scenarios in Figure 3 found on page 6. However, the compression requirements do not differ between scenarios.

Comment 3: In Section 2, the Asteroid Image Simulator is the key tool for generating simulated data. It is recommended to supplement the detailed descriptions of this tool and add a flowchart illustrating the process of simulated data generation.

Response 3: A more detailed description would be available in a more general article about the ASPECT camera unit in the Hera mission. This article by A. Näsilä et al. has been submitted to Space Science Reviews to be published in a special issue focusing on the Hera mission, but unfortunately, the article is still under review. However, we added a flowchart about the simulations in Section 2, page 5, Figure 3.

Comment 4: Section 4, Discussion: The analysis in this section is too broad. It is recommended to conduct a more quantitative comparison of the performance of these two methods, and further explore the boundary conditions of each compression method used in different scenario.

Response 4: We agree and have added more specific data to the discussion accordingly and described both algorithms' behavior in both boundary conditions (high and low noise). The entirety of the revisioned subsection 4.1. "Interpretation of Results" can be found spanning pages 14 and 15. 

Comment 5: Spectral information is extremely crucial for hyperspectral data. Could you provide a more detailed analysis about the ability of these two compression methods to preserve and reconstruct spectral information?

Response 5: While we agree that spectral information and its preservation are vital, the manuscript only compares lossless compression methods. Therefore, the decompressed data is always identical to the original and the spectral information is fully recovered. We did not investigate lossy compression options.

Reviewer 3 Report

Comments and Suggestions for Authors

1.The two image compression algorithms CCSDS 122.0 and JPEG 2000 are not briefly introduced in the abstract.

2. What do the previous findings in the abstract refer to ?

3.What are the innovations of the manuscript?

4.The notes are too brief and there is no clear description of the keywords in the picture.

5 The literature sources and references of noise filtering methods are not given in this paper.

Author Response

We thank you for spending your time carefully reviewing the manuscript and for your opinions regarding the science and presentation of the material.

Comment 1: The two image compression algorithms CCSDS 122.0 and JPEG 2000 are not briefly introduced in the abstract.

Response 1: We agree with this comment and have now specified that the algorithms are wavelet-based as well as the origins of their implementations in the abstract. The updates can be found on lines 3-6.
"This study compared two wavelet-based image compression algorithms, CCSDS 122.0 and JPEG 2000, used in the European Space Agency Comet Interceptor and Hera missions, respectively, in varying scenarios. The JPEG 2000 implementation was sourced from the JasPer library, whereas a custom implementation was written for CCSDS 122.0."

Comment 2: What do the previous findings in the abstract refer to ?

Response 2: We changed the wording of the abstract to be clear to the reader. Previous findings are now only mentioned in the discussion section.

Comment 3: What are the innovations of the manuscript?

Response 3: The most significant achievement of this study is the head-on quantitative comparison between commercial (JasPer) and space technology-based (Our proprietary CCSDS 122.0) implementations of the wavelet transform compression algorithms, including various noise conditions, pre-processing (noise filtering, diff. encoding) in terms of compression efficiency and computing costs.

Comment 4: The notes are too brief and there is no clear description of the keywords in the picture.

Response 4: We agree and have detailed the figure captions further and included a detailed legend in each figure.

Comment 5: The literature sources and references of noise filtering methods are not given in this paper.

Response 5: Thank you for pointing this out. We agree with this comment and have added sources accordingly. The updated citations can be found in section 2.5 "Noise Filtering" spanning pages 8 and 9. 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The author made extensive revisions to the article and answered all of my concerns. I have not other questions.

Reviewer 3 Report

Comments and Suggestions for Authors

My concenrns have been addressed. It can be accepted with minor language polishing from my point.

Comments on the Quality of English Language

The language can be further improved.