dose anybody use the openjpeg's j2k encoder(image_to_j2k)
and decoder (j2k_to_image) over 16 bit color image ?
These days, I use image_to_j2k to compression some 16 bit color tiff
image, and I found that
although I set the -r( for compression ratio) or -q (for quanlity)
parameters, but the it looks
like that the compressed images's quanlity is not according to the
parapeters, if
I compute the PSNR, I found that for -r from 2 to 50, the result is
almost the same,
do I need some other setting for 16 bit images ? ( I use the default
setting except the
-r or -q options.)
Thanks a lot !
While I'm not clear on the openjpeg implementation, you should be aware
that when you view 16bit images all that is first rendered down to 8bpp,
so any effects on the lower bitplanes are lost in first place. It also
means that most of the data in the lower bitplanes is just noise, which
is hard to compress, and which you cannot (visually, i.e. on the screen)
see. This means that the compression ratio are usually seemingly
"higher" for a given quality than you might expect. I don't know whether
openjpeg offers that, but a "compress to a PSNR" switch would then be
more valuable than giving a compression ratio (our software has that).
So yes, the world of 16bpp is definitely different.
So long,
Thomas