A modified hamming distance measure for quick detection of dissimilar binary images

Abstract

Given two dissimilar binary images we are interested in detecting their dissimilarity quickly. In our previous work, we developed a Probabilistic Model for Matching (PMM) that can be used to detect distinct-dissimilar binary images quickly with high confidence; e.g., by checking as few as five points with 93% confidence. In this paper we develop a similarity measure to be used with PMM. The similarity measure is a modification of the Hamming distance that measures the concurrence between two images. We show using real images, that by using this similarity measure, PMM can be extended to quasi-similar (non-distinct-dissimilar) images.