基于量子遗传和无迹粒子滤波的人体运动跟踪

提出一种人体运动跟踪算法,用于解决多关节人体运动跟踪问题.由于无迹粒子滤波存在样本贫化现象,因而对目标运动估计产生影响,尤其估计模型为复杂的马尔可夫链的时域问题的影响尤为严重.通过分析产生该现象的原因,在无迹粒子滤波中引入量子遗传算法:一方面,增加样本集的多样性而缓解样本贫化现象的影响;另一方面,改善其估计、跟踪能力并有效缩短了计算时间.实验结果表明,所提出算法很好地减轻了样本贫化现象对无迹粒子滤波的影响,并提高了多关节人体运动跟踪的准确性,跟踪结果令人满意.

Human Motion Tracking Based on Quantum Genetic and Unscented Particle Filter

A new algorithm was proposed to solve the movement tracking of multi-articulated human model. Sample impoverishment phenomenon exists in the unscented particle filter, which will affect the objects motion estimation, especially in the cases to estimate the model that is the time domain problem of complicated Markov chains. Based on the analysis of the cause of sample impoverishment, quantum genetic algorithm was introduced into the unscented particle filter to solve the problem. First of all, sample impoverishment was relieved by increasing the diversity of samples set, then the estimation and the tracking ability were ameliorated, at the same time, the computation time was efficiently reduced. Experimental results demonstrate that the proposed algorithm can alleviate the effect of the sample impoverishment phenomenon for the unscented particle filter and improve the tracking veracity of the articulated human movement. The tracking results are satisfactory.