A low power implementation for the motion estimation processor
Jer Min Jou, Yeu Horng Shiau, Bu Ren Zheng · 2004
This paper presents a low power video motion estimation processor. The combination of motion fuzzy prediction and modified four-step search algorithm (4SS) gives faster and more correct searching results. With the analysis of the relationship between the consecutive search iterations, some of the searching operations are saved which enhance the computing efficiency and reduce power consumption. Experimental results demonstrate the higher performance in both PNSR and computation complexity compared to other systems.