Results 1 to 5 of 5

Thread: Seeking image encoding benchmark

  1. #1
    Member
    Join Date
    May 2014
    Location
    Canada
    Posts
    141
    Thanks
    72
    Thanked 21 Times in 12 Posts

    Seeking image encoding benchmark

    I have been working for the past year and a half on an accelerated JPEG 2000 codec.
    The encoder will be complete in a few weeks.

    Now, I would like to benchmark the speed and compression performance of the codec; is there a
    an existing benchmark I can use, or do I have to roll my own? Because, its funny, I see people claiming
    various frame rates for their codecs, but they don't mention what types of images they tested on.
    So, these are meaningless; a series of all-black images are going to be fast, random noise not so much.

    I would like to test on a large collection of both single and multiple image sets, RGB and mono, at various frame sizes,
    lossy and lossless, and measure speed, compression ratio, and PSNR (for lossy).

    Thanks.

  2. #2
    Member
    Join Date
    Jun 2015
    Location
    Switzerland
    Posts
    900
    Thanks
    246
    Thanked 326 Times in 199 Posts
    In past studies JPEG2000 compared to JPEG was typically better in a PSNR study, but a little worse in human eval. Consider including a human eval if you can.

  3. Thanks (2):

    boxerab (30th October 2015),Cyan (30th October 2015)

  4. #3
    Member
    Join Date
    May 2014
    Location
    Canada
    Posts
    141
    Thanks
    72
    Thanked 21 Times in 12 Posts
    Thanks, Jyrki. I'm not interested in comparing JPEG 2000 to other codecs, but to compare different implementations of JPEG 2000.

    I think a small software package would be very useful here: combine a large collection of public domain sample images with a cross-platform viewer that runs
    automatic analysis of a codec, and then presents results to the user, plus the option of comparing any two images side by side (or ability to toggle the two
    images in the same viewing frame, to detect differences). In the case of JPEG 2000, there is already an open source tool to analyse conformance of the meta data
    to the standard (JPylzer), but nothing to analyse image quality.

  5. #4
    Member
    Join Date
    Sep 2007
    Location
    Denmark
    Posts
    926
    Thanks
    58
    Thanked 116 Times in 93 Posts
    if you mean pictures source to use for making benchmarks then i use these two collections. THe haven been sued multiple times in other benchmarks as well

    http://imagecompression.info/test_images/
    http://r0k.us/graphics/kodak/

  6. #5
    Member
    Join Date
    May 2014
    Location
    Canada
    Posts
    141
    Thanks
    72
    Thanked 21 Times in 12 Posts
    Thanks, Sven. These will be useful. I think a general framework for benchmarking, auto analysis, and human inspection would be useful.

  7. Thanks:

    Jyrki Alakuijala (30th October 2015)

Similar Threads

  1. Replies: 38
    Last Post: 27th April 2016, 18:01
  2. Replies: 4
    Last Post: 11th September 2015, 22:47
  3. Seeking Right click md5 & sha calc/verified for win64
    By SvenBent in forum The Off-Topic Lounge
    Replies: 4
    Last Post: 12th January 2015, 21:04
  4. Replies: 33
    Last Post: 27th August 2011, 05:13
  5. Graphic/Image/Modeling Benchmark
    By Simon Berger in forum Data Compression
    Replies: 23
    Last Post: 24th May 2009, 17:50

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •