基于方向自适应十字搜索的快速块匹配运动估计算法

Novel Oriented Adaptive Cross Search for Fast Block Matching Motion Estimation

提出了一种方向自适应十字搜索算法,通过自适应地使用小十字模板、大十字模板和四种方向的T形模板,有效地减少了搜索点数,提高了搜索速度。实验结果表明,该算法在保持与菱形搜索(DS)、正方形—菱形搜索(SDS)、十字—菱形搜索(CDS)和小十字—菱形搜索(SCDS)四种算法相同搜索精度的同时,速度上比DS、SDS、CDS和SCDS算法分别提高了74.65%、39.78%、42.44%和7.84%。

A novel Oriented Adaptive Cross Search （OACS） algorithm was proposed, where small cross, large cross and Tshape search-patterns were utilized adaptively, The proposed algorithm not only might reduce the number of searching candidates, but also could reach the global or close-to-global optimization. Experimental results had shown that OACS algorithm could speed up 74.65% , 39.78% , 42.44% , and 7, 84% over the Diamond Search（DS） , Souare-DS（SDS） , Cross-Diamond Search（CDS） , and Small-CDS（SCDS） algorithms, respectively, and it could maintain the similar search accuracy.