Adaptive multi-feature tracking in particle swarm optimization based particle filter framework  

作　　者：Miaohui Zhang Ming Xin Jie Yang 

机构地区：[1]Department of Automation, Shanghai Jiao Tong University, and Key Laboratory of System Control and Information Processing, Ministry of Education of China, Shanghai 200240, P. R. China [2]Institute of Image Processing and Pattern Recognition, Henan University, Kaifeng 475001, P. R. China

出　　处：《Journal of Systems Engineering and Electronics》2012年第5期775-783,共9页系统工程与电子技术（英文版）

基　　金：supported by the Chinese Ministry of Science and Intergovernmental Cooperation Project (2009DFA12870);the National Science Foundation of China (60974062,60972119)

摘　　要：This paper proposes a particle swarm optimization（PSO） based particle filter（PF） tracking framework,the embedded PSO makes particles move toward the high likelihood area to find the optimal position in the state transition stage,and simultaneously incorporates the newest observations into the proposal distribution in the update stage.In the proposed approach,likelihood measure functions involving multiple features are presented to enhance the performance of model fitting.Furthermore,the multi-feature weights are self-adaptively adjusted by a PSO algorithm throughout the tracking process.There are three main contributions.Firstly,the PSO algorithm is fused into the PF framework,which can efficiently alleviate the particles degeneracy phenomenon.Secondly,an effective convergence criterion for the PSO algorithm is explored,which can avoid particles getting stuck in local minima and maintain a greater particle diversity.Finally,a multi-feature weight self-adjusting strategy is proposed,which can significantly improve the tracking robustness and accuracy.Experiments performed on several challenging public video sequences demonstrate that the proposed tracking approach achieves a considerable performance.This paper proposes a particle swarm optimization（PSO） based particle filter（PF） tracking framework,the embedded PSO makes particles move toward the high likelihood area to find the optimal position in the state transition stage,and simultaneously incorporates the newest observations into the proposal distribution in the update stage.In the proposed approach,likelihood measure functions involving multiple features are presented to enhance the performance of model fitting.Furthermore,the multi-feature weights are self-adaptively adjusted by a PSO algorithm throughout the tracking process.There are three main contributions.Firstly,the PSO algorithm is fused into the PF framework,which can efficiently alleviate the particles degeneracy phenomenon.Secondly,an effective convergence criterion for the PSO algorithm is explored,which can avoid particles getting stuck in local minima and maintain a greater particle diversity.Finally,a multi-feature weight self-adjusting strategy is proposed,which can significantly improve the tracking robustness and accuracy.Experiments performed on several challenging public video sequences demonstrate that the proposed tracking approach achieves a considerable performance.

关 键 词：particle filter particle swarm optimization adaptive weight adjustment visual tracking 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] TP391.41[自动化与计算机技术—计算机科学与技术]