Hierarchical motion estimation algorithm based on pyramidal sucessive elimination

Abstract

In this paper, wepropose a fast motion estimation algorithm using pyramid hierarchy. Hierarchical Minkowski inequality is adopted to reduce the complexity of Mean Absolute Difference(MAD) computations. A top-town procedure is developed to search motion vectors hierarchically so that the total number of matching points is reduced drastically. Thespatial/temporal correlation is also considered to further speed up the searching process.Experimental results show that the proposed algorithm can achieve high com putation efficiency while maintaining comparable PSNR performance.