摘要
利用增强位平面的二值特性和运动矢量场自适应思想,提出了一种新的自适应二值运动估计算法。算法在图像的位平面中完成运动矢量的搜索;匹配时对静止块进行判定直接中止搜索,同时采用了起始点预测和运动类型判定,自适应使用LDSP和SDSP模板进行搜索。实验结果表明:该算法的搜索速度优于绝大多数现有的运动估计算法,且预测精度接近于FS算法。
Motion estimation is a key technique in video coding. In this paper, a novel adaptive binary motion estimation algorithm is proposed, which combines binary ehanced bit-plane and motion vector field. The searching procedure is processed in the 3rd bit-plane in place of the original gray image, the algorithm also utilizes effective half-way stop criteria, prediction of initial searching point, classification of macroblocks and 2 different searching patterns＂ LDSP and SDSP. Experimental results show that the proposed algorithm has a significant speed-up over most other algorithms and could obtain almost the same PSNR performace as the FS algorithm.
出处
《电路与系统学报》
CSCD
北大核心
2007年第1期104-108,共5页
Journal of Circuits and Systems
基金
国家自然科学基金资助项目(69975015)
教育部优秀青年教师资助计划项目(2003)
