From: Abhishek D. <the...@gm...> - 2009-03-01 14:34:31
|
Hi Bibek, I am not aware about any rejection techniques. However, if you pass those windows (containing wall) to lda or kepenekci face recognition module, then we can easily tell you if it has features of a human face or not. LDA and Kepenekci modules perform the recognition process at very high speed (its speed will increase even further is we use OpenMP to parallelize stuffs). So for now, you can leave it to FR modules to reject non-faces (for now only). bye On Sun, Mar 1, 2009 at 3:53 PM, Bibek Shrestha <bib...@gm...>wrote: > Neural networks detection is working properly right now. > One problem that has arized is.. > > Since we are doing histogram equalization in the image, this causes a lot > of noise to smoother image. > > For example say an image subwindow contains very small variations around > range of 240 to 250 only (consider a wall), such image after histogram > equilization will have a lot of noise after histogram equilization and > produce face like structure. > > i have been searching for possible rejection techniques to reject these > windows beforehand, by checking the pixel variation but still haven't found > a permanent solution > > Anjan, you once suggested me about such similar technique, do you have any > idea about it? > > If any one knows any technique, it would greatly speed up the filtering > process. > > Cheers > > -- > Bibek Shrestha > bibekshrestha at gmail dot com > > > ------------------------------------------------------------------------------ > Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, > CA > -OSBC tackles the biggest issue in open source: Open Sourcing the > Enterprise > -Strategies to boost innovation and cut costs with open source > participation > -Receive a $600 discount off the registration fee with the source code: > SFAD > http://p.sf.net/sfu/XcvMzF8H > _______________________________________________ > Rtftr-devel mailing list > Rtf...@li... > https://lists.sourceforge.net/lists/listinfo/rtftr-devel > > -- Abhishek http://adutta.np.googlepages.com |