Systematic way to compare 2 images

[watashi]

It is not known beforehand whether two images have to translate and rotate before comparison.

I tried bunwarp, which looks quite good but just don't know why it will be overwhelmed in some occasions. Can SURF do the job?

The program has to first align and anchor the images (registration?) and then perform pixel by pixel comparison (my images are just black-and-white) and then tell the difference (so identical images should give 0)

Any advice (even negative ones) is welcomed.

[komal_asrani]

CONTENTS DELETED

The author has deleted this message.

[watashi]

komal_asrani wrote

Hello

I would suggest that the pixel data for the image can be stored in a regular pattern as it is generated. Our algorithm need to match the whole  array to find the first match. The comparison would start from that particular match.This would ensure rotation invariant.

With regards

Komal Asrani

The images are generated by microscope so I can't control the process. However, do you mean you have an algorithm that can do the job?