A new adaptive fast motion estimation algorithm based on local motion similarity degree (LMSD)  

作　　者：LIU Long HAN Chongzhao BAI Yan 

机构地区：[1]School of Electronics ＆ Information Engineering, Xi＇an Jiaotong University, Xi＇an 710049, China

出　　处：《Progress in Natural Science:Materials International》2005年第12期1126-1134,共9页自然科学进展·国际材料（英文版）

基　　金：Supported by the Major State Basic Research Development Program of China (Grant No. 2001CB309403)

摘　　要：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 I,MSD; then different proposed search schemes are adopted for motion estimation. The experimental results show that the proposed algorithm has a significant computational speedup compared with MVFAST and PMVFAST algorithms, and offers a similar, even better performance.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 show that the proposed algorithm has a significant computational speedup compared with MVFAST and PMVFAST algorithms, and offers a similar, even better performance.

关 键 词：motion estimation motion vector video coding local motion 

分 类 号：TN919.81[电子电信—通信与信息系统]