The paper indicates this heuristic works in 'most' cases. :-)
It seems to me that the problem caused in this example is because
the 8x8 window covers other interesting features and is too large
for this particular case.
Far be it for me to suggest modifying their algorithm... But...
- If you used variable sized centred windows (4,6,8,10) and
calculated the weight for each window. There will be a window size
where the weight flips. This might suggest an improvement to the
heursistic where the state is constant for two scales in a row.
- Another approach might be to perturb the center point of the
window by 1 position in all 8 directions and take the majority
weight in the 8x8 case. I did not calculate this - just a hunch.
But I think the first may work better...
Cheers...
On 6/15/2013 8:52 PM, Vinícius dos
Santos Oliveira wrote: