A New Fast Motion Estimation Algorithm
In the motion vector field adaptive search technique (MVFAST) and the predictive motion vector field adaptive search technique (PMVFAST), the size of the largest motion vector from the three adjacent blocks(left, top, top-right) is compared with the threshold to select different search scheme. But a suitable search center and search pattern will not be selected in the adaptive search technique when the adjacent motion vectors are not coherent in local region. This paper presents an efficient adaptive search algorithm. The motion vector variation degree (MVVD) is considered a reasonable factor for adaptive search selection. By the relationship between local motion similarity degree(LMSD) and the variation degree of motion vector(MVVD), the motion vectors are classified as three categories according to corresponding LMSD; then different proposed search schemes are adopted for motion estimation. The experimental results shows the proposed algorithm has a significant computational speedup compared with MVFAST and PMVFAST algorithms, but offers a similar, even better performance.
motion estimation motion vector local motion
Long Liu Ding Liu
The Faculty of Automation and Information Engineering,Xian University of Technology Research,Xian, 710049,P.R.China
国际会议
The International Colloquium on Onformation Fusion 2007(2007年国际信息融合研讨会)
西安
英文
318-323
2007-08-22（万方平台首次上网日期，不代表论文的发表时间）