Novel unsymmetrical dual cross-diamond search algorithm for fast block motion estimation
Bin Sun, Zhi Liu, Haixia Zhang · 2014
In block-matching motion estimation, the searching speed and quality depend highly on the search patterns with different shapes and sizes. In this work, we propose a novel unsymmetrical dual cross-diamond search (DCDS) algorithm, which employs unsymmetrical cross-search patterns in the initial steps and small diamond-search patterns in the subsequent steps for fast block motion estimation. The initial cross-search patterns are designed for horizontal and vertical search patterns according to the center-biased characteristics of the motion-vector distribution. The proposed DCDS algorithm employs the halfway-stop technique and identifies motion vectors with fewer search points than other algorithms while maintaining similar or even better quality. Experimental results show that the DCDS algorithm is robust and has a shorter search time than other popular fast block-matching algorithms with comparative distortions.