XPost: sci.image.processing   
   From: vittrup@ima.auc.dk   
      
   On Sun, 27 Jul 2003, Allan Bruce wrote:   
      
   >"Alex"  wrote in message   
   >>   
   >> I'm looking for robust and possibly fast algorithms for recognition of   
   >> 2D-shapes (i.e. finding shapes similar to the given one). It is very   
   >> desirable for the algorithm to be invariant to scaling and rotation,   
   >> as well as to occlusion and noise/small defects. As already mentioned,   
   >> color and texture don't matter, I'm really only interested in finding   
   >> contours.   
   >   
   >This sounds like a job for Neural Networks - if the patterns are small then   
   >a backprop should be able to train for this, otherwise look at a SOM.   
      
   Nope. It doesn't matter if you choose an ANN or a Baye's Classifier or a   
   simple threshold value, if you're using wrong shape descriptors!   
      
     // Michael   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)