Good blog. Just so long as folks don't think SSIM is a panacea. I've
read the original paper and some others. SSIM is useful but, like
PSNR, it has issues too. VQEG have been trying to find a good quality
metric. There are lots of complex ones (SSIM is relatively simple,
which is good). Certain metrics seem to work well with similar types
of codec but not well with different types. For example they may work
well for MPEG codecs but not for JPEG2000 (and vice versa). So choice
of metric can lead to bias. This is one reason why PSNR is still used.
Tim
At 10:21 03/03/2010, Burkhard Plaum wrote:
Hi,
Saeed Salehi schrieb:
> Dear experts,
>
> I would like to compare the PSNR performance of Dirac codec with that of
> the H.264 (using *IBBBIBBBI...* frame format i.e. GOP = 4). I am just
> wondering what does the parameter *wlt_depth* mean exactly and what is
> its relation to the GOP size? is it the number of wavelet decompositions
> (i.e. GOP = 2 ^ wlt_depth ?) or it is simply the length of the group of
> pictures (GOP = wlt_depth ?). In other words, shall i use *wlt_depth =
> 2* or *wlt_depth = 4* for a fair comparison?
For a fair comparison of different codecs you shouldn't use PSNR. See here:
http://hirntier.blogspot.com/2010/01/video-quality-characterization.html
Burkhard
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Schrodinger-devel mailing list
Schrodinger-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/schrodinger-devel
No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 8.5.435 / Virus Database: 271.1.1/2719 - Release Date:
03/02/10 19:34:00
Tim Borer
BBC Research & Development
BBC Centre House
56 Wood Lane
London W12 7SB
Mobile: 07745 108652
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Schrodinger-devel mailing list
Schrodinger-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/schrodinger-devel
