Suppose we have two images. Both images are the same. The only different between them is that one of the images has a small filled circle on center. I can extract and visualize HOG features like this example:
http://se.mathworks.com/help/vision/ref/extracthogfeatures.html
And then calculate the Euclidean distance between two images by using hog features. In this case all hog features will be zero except the parts which related to filled circle area. BUT I don’t know how can I find the position of filled circle (the distinct area) and highlight it.
Hint: for example, for detecting people we can use this method:
I = insertObjectAnnotation(I,'rectangle',bboxes,scores);
http://se.mathworks.com/help/vision/ref/vision.peopledetector-class.html
Just substract two matrices, then find the bounding box on the result.