how to improve efficiency of image matching

i am doing project and in that i want to match a real time image with a reference image, for matching i am using histogram equilisation and after that substracting both the images, problem is that i am not getting a efficient result after applying my code for example: if the matching should be 80% i am oly getting 58% at max and not more than that so what should i do to have more efficient result.

following is my matching code:

a=histeq(a);
b=histeq(b);
z=imsubtract(a,b);
total_data = numel(z)
 matched_data = numel(z) - nnz(z)
numZeros = sum(sum(z==0));
total_matched_percentage = (matched_data/total_data)*100;

in the above code can we get number of '1' with number of '0' then this can improve our efficiency.

Answer 1

[Image Analyst]

It depends on what metric would be useful to you. Perhaps you just want to subtract the images and look at the area fraction of different pixels. Or maybe you want something more high level, like this from the Computer Vision System Toolbox. http://www.mathworks.com/products/computer-vision/features.html#feature-detection%2C-extraction%2C-and-matching