[Sls-mhdc] Entropy Coder Bit Rates (MHDC-A-1604-7)

[Kiely, Aaron B (332B)]

Dear Working Group,

Lucana attempted to send the message below to the Working Group, but the CCSDS mailing list seems to have discarded it.

Aaron




> ---------- Forwarded message ----------
> From: Lucana Santos <lsfalcon at iuma.ulpgc.es>
> Date: Tue, Sep 6, 2016 at 9:03 PM
> Subject: Re: [Sls-mhdc] Entropy Coder Bit Rates (MHDC-A-1604-7)
> To: "sls-mhdc at mailman.ccsds.org" <sls-mhdc at mailman.ccsds.org>
> 
> 
> Dear all,
> 
> I just uploaded a new xls sheet that contains the results that were
> missing due to corrupt files. In particular, the results that were
> marked as N/A in the _v2 version have been replaced by the correct
> values in the _v3 version. Additionally, I incuded a version of the
> same xls that just contains the data (no graphs and no intermediate
> computations). The file that only contains the data is the _v3_data.
> For the sake of completeness, I added also the results that Aaron
> included and were intially missing.
> 
> Since I imported the data to Matlab, I added also the file
> CCSDS_123_1_rates.rar in a folder called /matlab. Here you will find
> .mat files and a small script that reads just the rates from the
> files. I thought it could be useful for the group to have them in this
> format. In any case, I think Aaron's comprehensive report already
> offers enough information in order to reach conclusions about the
> performance of the entropy coders.
> 
> Kind regards,
> 
> Lucana
> 
> On Fri, Aug 26, 2016 at 12:22 AM, Kiely, Aaron B (332B)
> <aaron.b.kiely at jpl.nasa.gov> wrote:
>> Dear Working Group,
>> 
>> Here’s an update on entropy coder bit rate results for CCSDS-123.1-B proposals.
>> 
>> First, some of the zipped files of images of mapped quantizer indices were corrupted and could not be unzipped, and a couple of the image files had the wrong file names. My thanks to Enrico for spotting this issue, and my apologies to everyone for these mistakes. Because of the error, there are some omissions in the ESA reported results for a few images at a few values of max_err. I have uploaded corrected versions of the zipped folders that had problems. I believe that all of the zipped data are good now.
>> 
>> A couple of the range coding bit rates in the spreadsheet were incorrect. This has been fixed, and Lucana has uploaded an updated version of the spreadsheet with the corrections (the updated version has “_v2” in the filename).
>> 
>> I have uploaded an additional spreadsheet that repeats the submitted results (but in a different arrangement), but more significantly, includes some additional results. Most notably, results are included for the existing sample-adaptive (GPO2) and block-adaptive (Rice) entropy coders already standardized under CCSDS-123.0-B-1. The spreadsheet is here:
>>        CWE Private/123.1-B/Entropy Coders Comparison/EntropyCoders_more_data.xls
>> 
>> And finally, I have done some analysis of the different entropy coders and performance results and produced some summary slides which I have uploaded here:
>>        CWE Private/123.1-B/Entropy Coders Comparison/EntropyCoding_Analysis_Kiely2016-08-25.pdf
>> I welcome any feedback.
>> 
>> Action item MHDC-A-1604-7(e) calls for us to discuss whether the bit rate results provide enough information for us to make an entropy coder selection. My opinion is that the bit rates reported all represent acceptable performance for a new entropy coding standard, and I’d like to learn more about complexity before attempting to select an entropy coder.
>> 
>> Regards,
>> Aaron
>> 
>> _______________________________________________
>> SLS-MHDC mailing list
>> SLS-MHDC at mailman.ccsds.org
>> https://mailman.ccsds.org/cgi-bin/mailman/listinfo/sls-mhdc