Studies on digital image matching using ordered hierarchical technique

This paper presents an interesting approach to the image feature matching problem. Ordered hierarchical search algorithm is developed for the purpose of matching a template (M x M) using first order statistical features with overlapping partitioned pixel blocks of same dimension over an N x N, N > M, digital image. Mean, variance, skewness, kurtosis and energy features are chosen for comparison between the template block and test block. The order of matching, done judiciously, reduces the number of computations, thus reducing the computation time to a large extend. Computational analysis is done on two different images supporting the above facts.