[Octave-bug-tracker] [bug #46570] Image package: wrong result with edge

[Avinoam Kalma]

Follow-up Comment #4, bug #46570 (project octave):



>> im = ones(5,5);
>> im(3,3) = 0;
>> h1 = fspecial("sobel");
>> h3 = h1';

>> strength1 = sqrt( conv2(im, h1, "same").^2 + ...
                 conv2(im, h3, "same").^2 )
strength1 =

   4.24264   4.00000   4.00000   4.00000   4.24264
   4.00000   1.41421   2.00000   1.41421   4.00000
   4.00000   2.00000   0.00000   2.00000   4.00000
   4.00000   1.41421   2.00000   1.41421   4.00000
   4.24264   4.00000   4.00000   4.00000   4.24264

>> strength2 = sqrt( imfilter(im, h1, "replicate").^2 + ...
                 imfilter(im, h3, "replicate").^2 )
strength2 =

   0.00000   0.00000   0.00000   0.00000   0.00000
   0.00000   1.41421   2.00000   1.41421   0.00000
   0.00000   2.00000   0.00000   2.00000   0.00000
   0.00000   1.41421   2.00000   1.41421   0.00000
   0.00000   0.00000   0.00000   0.00000   0.00000



so conv2(im, h1, "same") gives spurious edges on the borders while
imfilter(im, h1, "replicate") doesn't. 
So it seems that imfilter is better than conv2.

imfilter with "replicate" has zero edges on the borders, which also seems to
be good. 
What do you think?



    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?46570>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/