基于遗传进化策略的粒子滤波视频目标跟踪  

Video Object Tracking Based on Particle Filter with Genetic Evolution Strategy

作　　者：席涛[1,2] 张胜修[1] 原魁[2] 颜诗源[1] 

机构地区：[1]第二炮兵工程学院,西安710025 [2]中国科学院自动化研究所,北京100190

出　　处：《光电工程》2009年第3期28-32,共5页Opto-Electronic Engineering

摘　　要：粒子退化问题是影响基于粒子滤波视觉跟踪性能的一个重要因素,为克服这种缺陷,本文将遗传进化策略引入粒子滤波跟踪算法,利用遗传算法的选择策略,根据预定的似然阈值迭代选择每代粒子中次优个体,然后对未选中的粒子实施交叉、变异操作以获得粒子的多样性,从而有效解决了粒子的退化问题。另外,针对跟踪中目标表观变化的问题,本文提出的跟踪算法采用了多模板自适应更新技术以确保跟踪的准确性。实验结果表明,本文提出的跟踪算法能有效地跟踪室内运动目标,并对光照变化、目标姿态变化具有良好的鲁棒性。Particle degeneration distinctly jeopardizes the visual tracking performance in particle filter. In order to deal with this shortcoming, an improved particle filter with genetic evolution strategy is addressed. In the proposed algorithm, the degeneracy problem is solved by applying the selection operator of genetic algorithm to choose the suboptimal samples iteratively based on the predefined likelihood threshold, and then the crossover and mutation operation are implemented to the samples which are not selected, so the diversity of the particles is maintained. Furthermore, considering the object appearance changes, the multi-template is employed to adaptively update the appearance of tracked object for keeping the accuracy in the scenario of tracking. The experiment results show that the proposed visual tracking algorithm can effectively track the moving object in the real-time indoor video sequences and is robust to the illumination and pose variations.

关 键 词：目标跟踪 粒子滤波 遗传进化 模板更新 跟踪算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]