Motion estimation using variable size block matching with cross square search pattern

Abstract

In recent decades, motion estimation is a major concern in video coding applications. Among all existing techniques, block matching is an emerging technique in motion estimation, because of its simplicity and effectiveness for both hardware and software implementations. In addition, the block matching technique helps in choosing the motion vectors for every macro block instead of utilizing a motion vector for every pixel. In this research article, Variable Size Block Matching (VSBM) with cross square search pattern is proposed for motion estimation in the videos. Compared to other block matching methods, the VSBM results in a low motion compensation error and also significantly speed up the motion estimation process. For reducing the illumination change effect in the video frame, cross square search pattern is developed that prevents macro blocks from over-splitting. In this work, the proposed algorithm performance is analysed by using some real time videos like container, Akiyo, Stefan, kristen-sara, foreman, Johnny, old-town-cross, Carphone, football and crowd-run. From the analysis, the proposed algorithm results in maximum of 14.9 dB improvement in peak signal to noise ratio value as compared to existing algorithms; dynamic pattern (square, diamond and hexagon) and harmony search with block matching.