normalized cross correlation template matching

Instead of just sliding the pattern over all the possible locations on the image we also consider all the possible rotations on each and every location. So we take the template and slide it over the whole image, calculate the normalized cross-correlation coefficient for each possible location on the image and consider the location having the highest normalized correlation coefficient as our matching region.Īs mentioned at the beginning of the post, we can take rotation into account as well.

The straightforward and simple answer is… Yes, you guessed it right, everywhere! Now the question is where to look in the image? Exhaustive Search Okay, we have all the building blocks for image-template matching. Instead, the cross-correlation function is used for measuring similarity giving the method its name. Hence, in the cross-correlation template matching approach, the above techniques of similarity matching are not employed. The same intensity values of the template and image patch are not guaranteed even though they are matching. If we simply add 1 to all the intensity values of our template then the output would be completely different for the above measures of similarity.īut as defined in our goal, we want our template matching to be invariant to the changes in brightness and contrast of template and image. The issue with the above-mentioned methods is that they are not invariant under changes in brightness or contrast. The typical ways of measuring similarity are: Now that we know the goal of template matching, the next thing we need to figure out is how to quantify similarity, how to measure that one location is a better match than the other? Quantifying the results of template matching

Normalized cross correlation template matching