Journals Proceedings

Abstract

Many fast block matching algorithms have been proposed to reduce computational complexity in motion estimation process. Previously developed hexagonal algorithms focus on improvement of either low resolution coarse search or fine-resolution inner search. In this paper, proposed algorithm named Directional Hexagon-based Block Search (DHEXBS) improves both coarse and inner search. The proposed algorithm reduces the number of search points by exploiting the distortion information in the neighboring search points. Our experimental results show that the proposed algorithm substantially outperforms HEXBS in terms of number of search points while maintaining the Peak Signal-to-Noise ratio. (PSNR)