Comments on: Selective Mode filter in Matlab http://www.shawnlankton.com/2008/04/selective-mode-filter-in-matlab/ life, business, consulting, and computer vision Tue, 07 Feb 2012 12:43:37 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Shawn Lankton http://www.shawnlankton.com/2008/04/selective-mode-filter-in-matlab/comment-page-1/#comment-1340 Shawn Lankton Thu, 29 Apr 2010 18:49:38 +0000 http://www.shawnlankton.com/2008/04/selective-mode-filter-in-matlab/#comment-1340 <a href="#comment-1321" rel="nofollow">@amira </a> I did not use belief propagation here. The "selective mode filter" is a hack-y way to simulate belief propagation quickly with simple-to-write codes : ) @amira
I did not use belief propagation here. The “selective mode filter” is a hack-y way to simulate belief propagation quickly with simple-to-write codes : )

]]> By: amira http://www.shawnlankton.com/2008/04/selective-mode-filter-in-matlab/comment-page-1/#comment-1321 amira Fri, 23 Apr 2010 16:17:16 +0000 http://www.shawnlankton.com/2008/04/selective-mode-filter-in-matlab/#comment-1321 Hello Yawn, Great work, thanks for shared it. I am asking if you have used the belief propagation at the final step of your code. Hello Yawn,
Great work, thanks for shared it.
I am asking if you have used the belief propagation at the final step of your code.

]]> By: sergo http://www.shawnlankton.com/2008/04/selective-mode-filter-in-matlab/comment-page-1/#comment-312 sergo Tue, 10 Jun 2008 08:57:24 +0000 http://www.shawnlankton.com/2008/04/selective-mode-filter-in-matlab/#comment-312 in my view "your" first implementation was much better than this ;-) Did you try Belief Propagation? in MEX it works just slightly slower than WTA+Median, but results are more reliable http://people.cs.uchicago.edu/~pff/bp/ (and then Median filter adds a bit more dB in PSNR) How do you think to incorporate temporal redundancy between video frames to improve estimate? in my view “your” first implementation was much better than this ;-)

Did you try Belief Propagation? in MEX it works just slightly slower than WTA+Median, but results are more reliable
http://people.cs.uchicago.edu/~pff/bp/
(and then Median filter adds a bit more dB in PSNR)

How do you think to incorporate temporal redundancy between video frames to improve estimate?

]]>