Re: RLE & RGB to YBR_FULL

On Mar 23, 9:58 pm, Chuck McCrobie <x...@xxxxx> wrote:
Part 5, Section 8.2.2 Note 2 states, in part,

"It is unusual to use an RGB colorspace for RLE compression, since no
advantage is taken of correlation between the red, green and blue
components (e.g. of luminance), and poor compression is achieved (note
however that the conversion from RGB to YBR FULL is itself lossy. A new
photometric interpretation may be proposed in the future which allows
lossless conversion from RGB and also results in better RLE compression
ratios)."

My question is:

Does this mean if I RLE encode an image and do color space conversion
from RGB to YBR_FULL that I must mark the composite as "Lossy"
(0028,2110) with "01" and make a new SOP Instance UID for the newly RLE
compressed composite?

Does color space conversion make for lossy compression? I understand
that JPEG 2000 has both YBR_ICT for lossy color space conversion and
YBR_RCT for lossless color space conversion. Unforunately, the same
does not apply to RLE does it?

Thanks,

Chuck

Sorry for replying to my own message, but I managed to find this:

http://groups.google.com/group/comp.protocols.dicom/browse_thread/thread/720276384fd28a22/2b1a28dbcd15f448?hl=en&lnk=gst&q=rle+color+conversion+lossy#2b1a28dbcd15f448

In short, it is "recommended" that conversion from YBR_* to RGB be
considered lossy and thus require a new SOP UID and setting the Lossy
and LossyCompressionRatio elements. I would imagine the converse is
true too - going from RGB to YBR_*

David Clunie writes:
Hi Don

Though the rule requiring a new SOP Instance UID is based
on changing the "interpretation", I would argue strongly
that whenever there is a non-lossless transformation of
the pixel data one should err on the side of changing the
UID, as in this case, even though the loss is small and
due to calculation error.

david

PS. I can't imagine anyone would ever use RLE and YBR for
XA images - surely only for color US.
.

Relevant Pages

  • Re: RLE & RGB to YBR_FULL
    ... from RGB to YBR_FULL that I must mark the composite as "Lossy" with "01" and make a new SOP Instance UID for the newly RLE compressed composite? ... that JPEG 2000 has both YBR_ICT for lossy color space conversion and YBR_RCT for lossless color space conversion. ... YBR_RCT is a transformation that has been designed specifically as ... at the expense of sacrificing a little bit of compression ...
    (comp.protocols.dicom)
  • Re: JPEG2000
    ... We are in need of a compression meganism that is capable of compressing ... use larger pixels/fewer bits per pixel. ... maybe huffman and some manner of filtering "could" get it a little higher, ... huffman coder has rle. ...
    (comp.compression)
  • Re: GetThumbnailImage increases file size
    ... ResizeAndSaveToAll = ResizedImg.Size.ToString is not nessesarly the ... That your orignial image was only 42k suggests it was using rle ... compression. ... together and the length of consequative identical pixel sequences is ...
    (microsoft.public.dotnet.framework.drawing)
  • Re: Fujitsu NetCOBOL for .NET
    ... With a 69-byte "text" file, I doubt you'd get much compression, if ... from RLE; plain ASCII text (which I assume is what we're dealing ... LZ77 doesn't require much table space. ... symbols in LZ77 output refer back to earlier data: ...
    (comp.lang.cobol)
  • Re: TIFF compression
    ... >I vaguely read packbits uses RLE encoding, though I am not sure whether the ... >RLEencoding scheme by the tiff encoder in GDI+. ... PackBits is an RLE compression method which works with any color ...
    (microsoft.public.win32.programmer.gdi)