关键词: 块匹配, 运动估计, 方向自适应, 局部采样, 预测

Abstract: In order to improve the searching accuracy of motion estimation, a new algorithm called Directional Adaptive Sampling Search (DASS) algorithm was proposed. According to the information of matching error in search patterns, the algorithm adopts two sorts of triangle search patterns adaptively. The strategy of local sampling search was employed for the sake of big-motion vector. Early termination strategy was adopted in the procedure of DASS so as to further speed up the search. The experimental results show that the proposed algorithm not only ensures the speed but also improves the accuracy of search with Peak Signal-to-Noise Ratio (PSNR) degradation of only 0.28dB on average.

Key words: block matching, motion estimation, directional adaption, local sampling, prediction