Novel Oriented Adaptive Cross Search for Fast Block Matching Motion Estimation

A novel Oriented Adaptive Cross Search(OACS) algorithm was proposed,where small cross,large cross and T-shape 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.