jitter_dist

 Calculates the minumum distance between two sets of possibly 'jittered' feature vectors.

 Normally a feature vector is a p element vector.  If there is jitter that means an
 object is described by r such p element vectors, taken from different 'views'. The
 distance between a jittered vector (r 1xp vectors) and a 1xp vector is simply taken as
 the minimum of the r distances.  

 If both desc1 and desc2 are jittered vectors, than a full search is not performed -
 instead the following is used to obtain the distance:
   D = min( jitter_dist(desc1(middle),desc2), jitter_dist(desc1,desc2(middle)) ).
 
 INPUTS
   desc1       - m x p x r1 array of m rxp jittered vectors (r1=1 ok) 
   desc1       - n x p x r2 array of m rxp jittered vectors (r2=1 ok) 
   dist_func   - [optional] distance function that regular non jittered vectors

 OUTPUTS
   D           - m x n array of distances

 EXAMPLE
   load testimages;
   A = double(squeeze(reshape( images, [], size(images,3) )))';
   figure(1); im( jitter_dist(A,A) ); 
   B = jitter_image(images,1,0,3,3);
   B = reshape( B, [], size(B,3), size(B,4) );
   B = double(squeeze(permute(B,[2 1 3])));
   figure(2); im( jitter_dist(B,B) ); 

 See also DIST_EUCLIDEAN