I want to find difference between two images.

One is a partial image of the other. Size is same. How get rid of black border. Tried XOR after binarizing(some processing applied in between but general idea is XOR). Fine results in image area but unwanted results on and outside border

[Walter Roberson]

What black border are you trying to get rid of?

I see a white border in the binary test image: are you referring to that ?

Answer 1

[Image Analyst]

You can get the bounding box of the small image by doing

[rows, columns] = find(binaryImage);
row1 = min(rows);
row2 = max(rows);
col1 = min(columns);
col2 = max(columns);
% Now crop the image to remove black frame
croppedImage = binaryImage(row1:row2, col1:col2);
% Resize it
binaryImage = imresize(binaryImage, size(otherImage));

Now you can compare the images by subtraction or xor() or whatever.

[faheem ifti]

Thanks sir

It works fine for the Test Image 2 against Reference image. But fails if the image is rotated as in image "Test Image 1". What can be an efficient method to restrict matching area of the two images only at the overlap of two images.

[Image Analyst]

Use a rotationally and scale invariant matching technique: Hu's image moments. See this link from the University of Arizona