Improvement on simplified UMHS algorithm

YANG Yu-hong,XU Xuan,Ji Xiao-yong
2007-01-01
Abstract:In this paper,based on the study of the principles of simplified UMHexagonS algorithm,some improved ways are proposed,such as using a new way instead of selecting only one predictor in the init selection step used by other predictive motion estimation that in the init predictor selection step,all of the predictors are used for the next step as different center points of different predictive areas.Furthermore,we use different search patterns for different motion type and so on.Simulation results show that with similar or even better rate-distortion performance,the proposed method decreases about 10%-40% of search points and reduces motion estimation time by about 6% to 20% on average when compare with those of simplified UMHexagonS algorithm.
What problem does this paper attempt to address?