Hexagonal Intersection-Based Inner Search to Accelerate Motion Estimation

摘要

The computational complexity of motion estimation (ME), increases proportionally with the number of search points. As a result, rapid ME techniques reduce complexity by using different search patterns. Of these techniques, hexagonal search (HS) with a small diamond search pattern (SDSP) significantly reduces complexity compared to other fast ME algorithms. The proposed hexagonal intersection search (HIS) algorithm improves the inner search of HS. The new algorithm cares the inner points near to the intersection of a hexagon rather than a SDSP or a full search. The proposed algorithm reduce 7-15% search points compared with HS. Compared with other popular algorithms, the HIS algorithm made the lowest number of average search points with neeotiable PSNR loss.