jpeg 2000 codec - DICOM

This is a discussion on jpeg 2000 codec - DICOM ; hi im writting a jpeg2k codec for DCMTK using jasper library, and i have been googling for a while and found nothing about it, i'd like to know if someone has implemented something alike with jasper. the main problem im ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: jpeg 2000 codec

  1. jpeg 2000 codec

    hi
    im writting a jpeg2k codec for DCMTK using jasper library, and i have
    been googling for a while and found nothing about it, i'd like to
    know if someone has implemented something alike with jasper. the main
    problem im facing is when i try to encode a dicom image already
    decompressed from any format to jpeg 2000, because jasper doesnt have
    an interface to encode from raw image, they always ask for an input
    format.

    ****BELOW FOR PEOPLE FAMILIARIZED WITH JASPER*******
    how can i create a "jas_image_t" or a jas_image_cmpt_t
    from( buffer,colorspace, height, width, sample, and all necessary
    info)

    when i finish i will make the code public under GPL
    thank for any help


  2. Re: jpeg 2000 codec

    > im writting a jpeg2k codec for DCMTK using jasper library, and i
    > have been googling for a while and found nothing about it, i'd
    > like to know if someone has implemented something alike with jasper.


    There is a ready-to-use implementation for the DCMTK :-)
    http://dicom.offis.de/dcmjp2k.php.en

    Regards,
    Joerg Riesmeier

  3. Re: jpeg 2000 codec

    On Apr 12, 1:26 pm, rodney....@gmail.com wrote:
    > hi
    > im writting a jpeg2k codec for DCMTK using jasper library, and i have
    > been googling for a while and found nothing about it, i'd like to
    > know if someone has implemented something alike with jasper. the main
    > problem im facing is when i try to encode a dicom image already
    > decompressed from any format to jpeg 2000, because jasper doesnt have
    > an interface to encode from raw image, they always ask for an input
    > format.
    >
    > ****BELOW FOR PEOPLE FAMILIARIZED WITH JASPER*******
    > how can i create a "jas_image_t" or a jas_image_cmpt_t
    > from( buffer,colorspace, height, width, sample, and all necessary
    > info)
    >
    > when i finish i will make the code public under GPL
    > thank for any help


    Hello,

    I have been posting multiple times about the issues with jasper
    (Marco from dcmtk also observed issues with this lib). Anyway if you
    *really* want to do it, hopefully this piece of code should help you
    (the last know time before the switch to openjpeg):

    http://cvs.creatis.insa-lyon.fr/view...?annotate=1.28

    We used it for a couple of months before discovering problems, but
    during that time it looked ok.

    Again you should really consider using OpenJPEG instead.

    2 cents
    Mathieu
    Ref:
    http://www.openjpeg.org


  4. Re: jpeg 2000 codec

    My JPEG2000 codec for DCMTK is using OpenJPEG with very good results,
    so I have to agree that OpenJPEG is the way to go. It performs much
    better than it used to and is much easier to dig around in. Note that
    OpenJPEG will use MCT on all 3-component images. That might come back
    to bite you for anything other than RGB. I made a patch to disable
    MCT for non-RGB images but it is quite dated and probably will not
    work with the new release of the library.

    Which brings up a question I have had for some time. Is there a way
    to determine what the original photometric interpretation was when you
    receive an image using YBR_RCT? In my J2K codec I assumed that they
    were always going to be RGB and I disable MCT for anything else. I
    did not take in to account any J2K headers marked as YUV or that were
    subsampled. I have yet to run across any such images. Are there any
    samples out there other than the ones by David Clunie?

    Is there anyone on the DCMTK team that would feel like I was stepping
    on toes if I posted my codecs somewhere? I currently have JPEG2000
    and JPEG-LS codecs which seem to work well. The JPEG-LS code uses a
    library written by me. I have yet to find any images that I have not
    been able to compress/decompress. Unfortunately there are not too
    many programs out there to test compatibility with.

    Colby Dillion


  5. Re: jpeg 2000 codec

    On Apr 12, 1:21 pm, "Colby Dillion" wrote:
    > My JPEG2000 codec for DCMTK is using OpenJPEG with very good results,
    > so I have to agree that OpenJPEG is the way to go. It performs much
    > better than it used to and is much easier to dig around in. Note that
    > OpenJPEG will use MCT on all 3-component images. That might come back
    > to bite you for anything other than RGB. I made a patch to disable
    > MCT for non-RGB images but it is quite dated and probably will not
    > work with the new release of the library.
    >
    > Which brings up a question I have had for some time. Is there a way
    > to determine what the original photometric interpretation was when you
    > receive an image using YBR_RCT? In my J2K codec I assumed that they
    > were always going to be RGB and I disable MCT for anything else. I
    > did not take in to account any J2K headers marked as YUV or that were
    > subsampled. I have yet to run across any such images. Are there any
    > samples out there other than the ones by David Clunie?
    >
    > Is there anyone on the DCMTK team that would feel like I was stepping
    > on toes if I posted my codecs somewhere? I currently have JPEG2000
    > and JPEG-LS codecs which seem to work well. The JPEG-LS code uses a
    > library written by me. I have yet to find any images that I have not
    > been able to compress/decompress. Unfortunately there are not too
    > many programs out there to test compatibility with.
    >
    > Colby Dillion


    Colby, Is your JPEG-LS codec a lib or a dll? I have been looking for
    something I could call from vb.net (or a JPEG lossless codex)
    Tom


+ Reply to Thread