[OctDev] Bug in imrotate.m

[Todd N. <to...@gm...>]

Hello,

I was looking at imrotate and tried it on a simple test matrix and got
the wrong result:

octave:1> a=[1,2,3;4,5,6;7,8,9]
a =

  1  2  3
  4  5  6
  7  8  9

octave:2> imrotate(a,270)
ans =

  0  7  4  1
  8  5  2  2
  9  6  3  0

I wrote a patch to handle the special cases where rotations are of
angles divisible by 90, it speeds up these rotations tremendously and
returns the correct result.

Before Patch:
octave:5> a=rand(1000);
octave:6> tic;imrotate(a,90);toc
ans = 16.821

After Patch:
octave:7> tic;imrotate(a,90);toc
ans = 0.37471


Thanks, 
Todd