Title :
Optimized H.264 motion estimation algorithm based on UMHexagonS
Author :
Liu Chen ; Jian-zhong Cao ; Li-nao Tang ; Ji-jiang Huang ; Hui-nan Guo ; Xiao-kun Dong
Author_Institution :
Xi´an Inst. of Opt. & Precision Mech., Xi´an, China
Abstract :
H.264 has adopted UMHexagonS algorithm as fast motion estimation algorithm for integer pixel formally, but this algorithm has some shortages such as search points are lots, quantity of operation is kind of big, costs much time, and so on. These need to be solved as soon as possible. This paper introduces and analyses UMHexagonS algorithm. In order to solve these problems, this paper brings forward classification strategy for search algorithm according to statistical properties of the motion vector prediction value, and improves the template based on original algorithm according to the characteristic of center bias of motion vector. The results of experiments show that the improved algorithm in this paper reduces the time of motion estimation by 15%~27%. At the meantime, it can ensure PSNR and rate basically unchanged.
Keywords :
image classification; optimisation; search problems; statistical analysis; video coding; UMHexagonS algorithm; center bias characteristic; classification strategy; integer pixel; motion vector prediction value; optimized H.264 motion estimation algorithm; search algorithm; statistical properties; Algorithm design and analysis; Classification algorithms; Diamonds; Motion estimation; Prediction algorithms; Support vector machine classification; Vectors; UMHexagonS; center bias; classification strategy; motion estimation; video compression;
Conference_Titel :
Intelligent Control and Information Processing (ICICIP), 2013 Fourth International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4673-6248-1
DOI :
10.1109/ICICIP.2013.6568063