Abstract
The same object seen in two different images can be geometrically and photometrically transformed. In this paper, we describe the method of matching interest points of the same object present in different images. With the knowledge of relative scales of the object, we compute local invariant descriptors for each of the detected interest points. This local invariant descriptor is used to match the interest points with the help of a hashing technique. The matching method could find a good number of correct matches, for different kinds of transformations in cluttered environments and partial occlusions.