[Octave-bug-tracker] [bug #50151] image package: normxcorr2 is not compatible with Matlab

[Hartmut]

Follow-up Comment #2, bug #50151 (project octave):

Here is another normxcorr2 result from Matlab and Octave, to compare:

I = [0 1 2 1 2;
     1 0 1 1 0;
     0 1 2 2 0;
     0 2 0 0 0;
     2 1 0 0 0] 
     
T = [0 1 1];

c = normxcorr2(T, I);
c(isinf(c))=0  # work around bug #50122

c_matlab = [ ...   # the Matlab result
          0    0.5000    0.8660    0.5000   -0.5000         0   -1.0000
     0.5000    0.5000   -0.5000    1.0000   -0.5000   -1.0000         0
          0    0.5000    0.8660    1.0000   -0.5000   -1.0000         0
          0    0.5000    0.5000   -1.0000         0         0         0
     0.5000    0.8660   -0.8660   -1.0000         0         0         0]

(abs(c - c_matlab)) > 1e-4

# result of this difference calcluation (in Octave):
#  1  1  0  0  0  1  0
#  0  1  0  0  0  1  1
#  1  1  0  0  0  1  1
#  1  1  0  0  0  1  1
#  0  1  0  0  0  1  1



This strengthens my suspicion that the significant differences are mainly
around the border pixels, where some kind of extrapolation needed to be done.



    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?50151>

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