Block Based Motion Estimation Approach Using Triple Successive Selection Stages Pedestal on Descriptors

Abstract triple successive selection stages of block matching approach based on block descriptors is introduced in this paper. Block descriptors serve the search and play the role of indexing. The searching scheme consist of three phases. Phase one include the calculations of block descriptor values. Phase two include triple successive selection stages, each stage use divergent block descriptor value to pick sub set of nominee seek points. Each selection stage has two filters. First filter based on the comparison between the descriptor value of the spots in the search region and the current block descriptor value. only the points that satisfy the selection condition passed that filter but the others obscured. Second filter regulate the number of points passed to the next selection stage. In phase three alteration quantity calculation performed to the most recent nominee seek spots to get the finest match and produce the movement information. For judgment purposes the well known Exhaustive Search (ES) and Three Step Search (TSS) algorithms are employed. The test results indicate that the approach matching block introduced quickly and efficiently, where its faster and more accurate than TSS as well as near the ES accuracy.