Optimizing Image Compression Ratio for Generating Highly Accurate Local Digital Terrain Models: Experimental Study for Martian Moons eXploration Mission
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe MMX project considers the landing site selection to be one of the most critical tasks. Due to the limited data downlink, landing sites must be selected only within a restricted area, which presents an extremely significant problem. To address this issue, this study examined how image data can be compressed by preparing simulated images in a laboratory to ensure that the safety and scientific value of operations are not compromised. The results showed that a Digital Terrain Model can be constructed with an error of approximately 40 cm if the compression ratio is 70% or less.
This achievement holds substantial implications as it suggests that the success of MMX may expand by potentially increasing the number of areas considered as potential landing sites. Furthermore, because this pioneering study can be adapted for future exploration by adjusting equipment parameters, it is expected to stimulate further research in this field. This study exemplifies meticulous experiment preparation and a well-described research methodology, making it highly valuable for future research in related areas.
Considering the points mentioned above, I strongly recommend the publication of this paper in this journal. It aligns with the journal's content and objectives and is likely to generate significant interest among readers.
Some minor comments below.
Line 63: There is no mention of observation of Phobos in the Emirates Mars mission [43].
Lines 113-124: The sentences between these lines are in the present tense, so it would be better to maintain consistency in tense with preceding text where appropriate.
Line 333: It is important to note that the influence of terrain roughness is more significant than the influence of the difference between integer and float compression methods. Additionally, it is crucial to quantitatively demonstrate that the error between the original surface and the Smooth surface is larger than that of the rough surface. Great work.
Line 391: in Figure 9, there is an overall blue circular ring in (c) and (d). I am curious whether this is dependent on the original elevation or if compression consistently produces such patterns. Perhaps a map displaying elevations would be helpful.
Lines 416-421: There is a wide variety of astronomical surfaces, including those with fewer boulders and fewer visible craters, as well as surfaces with distinct albedo patterns. In a way, this study focuses on the simplest and most extreme cases, and I hope that future research will build upon the foundation laid by this study to explore these diverse surface types.
Author Response
Please see the attached file, thank you.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper titled "Optimizing Image Compression Ratio for Generating Highly Accurate Local Digital Terrain Models: Experimental Study for Martian Moons eXploration Mission" analyze the impact of data compression to generate Digital Terrain Models, the conclusions of this study will be applied to the MMX mission of the JAXA. The main objective of this contribution is to figure out the maximum compression ratio that can be applied to the acquired images of the moons and still constructing high-quality DTMs to achieve safe landings. Although the paper looks very interesting in terms of space exploration, this has some lacks related with data compression. Though the authors can fix some issues of the paper, this reviewer suggests to the authors to look for an external institution with more experience on data compression.
Points be revised:
0) Line 162 says that "Image compression algorithms commonly perform two compression schemes: lossless and lossy compressions", which is true. However for science it is really interesting to control the maximum error between the original and reconstructed images. Thus also near-lossless compression techniques are employed. Introduce and analyze them experimentally should be interesting.
1) Line 171: The Consultative Committee for Space Data Systems is in charge of define standards for space and data information systems. These definitions are organized in books: blue books for the standard definition and green books for the technical reports. Along the document authors propose to uses CCSDS 120.0-B-1, which does not exists. The compression technique proposed is to use Discrete Wavelet Transform + Entropy Encoder, which is defined in CCSDS 122.0-B-2 (Image Data Compression). And CCSDS 120.1-G-3 corresponds to the latest version of the technical report of CCSDS 120.0-B-1.
Authors can fins al documents in the following link https://public.ccsds.org/Pubs/Forms/AllItems.aspx
2) Lines 177, 178 and 179 authors state:
"The integer-point DWT compression process is for lossless compression; it requires only integer arithmetic with lower implementation complexity, but it can also be used in lossy compression." This should be clarified: CCSDS 122.0-B-2 can store only a limited number of bytes defined by the user, thus the image data can be reconstructed partially. Because of the characteristic of the DWT and that the compression technique is embedded the image can be reconstructed progressively.
3) Line 186 is not correct. S_org --> original and S_comp --> compressed.
4) SSIM is a visual metric, whereas here the authors need to validate that the data is useful for science. Also include Peak Absolute Error (PAE) and Peak Signal Noise Ratio (PSNR) should be needed.
5) In line 204 and 205 is stated: "Leveraging high-end CPUs and GPUs (CPU of Core-i9 12900K and dual GPUs of GeForce RTX 3090), we reduced calculation times". This statement is interesting to compute the DTMs in ground stations. What is the consuming time of the process in the ground station?, how is this reduction in terms of computing time compared with only a single GPU? is not clear in the manuscript what happens after the DTM is created. This point cloud is returned to the spacecrfat? if this is correct: a) what is the time window to compute the DTMs and return the point clouds to the spacecraft? b) the point clouds to be returned are compressed?
6) Table 1 list the optimal parameters for obtaining high-precision local DTMs. How these parameters are obtained? reference? experiments to determine these parameters to obtain an error of at most 40 cm?
7) In section 3. Results only is evaluated CCSDS 122.0-B-2 using Integer and Floating Point Wavelet Transforms. Results for CCSDS 123.0-B-2 must be also considered, since it is the latest standard of image data compression and provide support for lossy compression also, and can manipulate gray images as CCSDS 120.0-B-1 does. Also results for JPEG2000 must be included, it might be considered the best standard in rate-distortion terms. Part-15 of JPEG2000 standard defines HTJ2K, which speeds-up JPEG 2000 by an order of magnitude at the expense of slightly reduced coding efficiency.
8) In line 276 is stated: "The processing speed of the CCSDS-120 with float-point DWT was approximately twice as fast as that of the integer-point DWT", however in line 177 and 178 authors say: "The integer-point DWT compression process is for lossless compression; it requires only integer arithmetic with lower implementation complexity, but it can also be used in lossy compression." Can you explain why Integer wavelet has a lower implementation complexity but is slower than Floating point path? In addition floating point includes a quantization stage.
9) Line 305 states: "Therefore, we recommend compression ratios of 70% or lower to reduce the data volume and simultaneously retain important scientific information." This a very interesting statement, and Figure 8 and 9 shows the DTMs error graphically for a compression ratio of 70%. However, this statement should be hold with more scientific results. For instance a plot in where in the horizontal axis there is the compression ratio and in the vertical axis the error of DTMs. With this plot the readers figure out easily the maximum compression ratio to be applied and the DTMs are still useful for safe landings.
10) Regarding the point 9) all compression is CCSDS 122.0-B-2 the best compression technique. Which is the behavior of DTMs error for different compression techniques at different compression ratios?
11) Regarding to the DTMs error, analyze the relationship between distortion metric PAE with the DTMs error, should be very interesting when near-lossless compression techniques are employed.
Author Response
Please see the attached file, thank you!
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsFrom the point of view of this reviewer, the title does not correspond with the contribution of the manuscript. There is no Optimization of any coding system to compute DTMs. This is just a validation of using CCSDS-122 in DTMs.
