【摘要】 通过对基于粒子滤波算法的运动目标跟踪技术进行研究,并针对粒子滤波算法的退化现象做出了两方面的调整。第一,对粒子滤波的重采样阶段做出了改进,在粒子上添加一个微小的高斯干扰,使得重采样的粒子分布发生变化,同时使采样枯竭得到了抑制;第二,经过一段时间的跟踪后,将跟踪目标重新初始化,继续跟踪,使得跟踪结果更加完善。通过自适应调整跟踪目标的窗口,使其大小改变,背景中的颜色尽量没有与跟踪目标相同的颜色。实验结果表明,这种改进过的粒子滤波算法能够在复杂的情况下进行跟踪,并且跟踪性能优于Meanshift方法。更多还原

【Abstract】 This article focus on the research of target tracking technology based on particle filtering algorithm, and two aspects are adjusted according to the particle filter algorithm degradation phenomenon. Firstly, resampling stage of particle filter is improved by plusing the tiny perturbations on the particles. And then resampling particle distribution is changed,and it inhibit s the sampling depletion. Secondly, it reinitializes tracking target after tracking a period of time, and tracking results will be more perfect. Meanwhile, tracking window’s size is adjusted. Experimental results show that the improved particle filter algorithm can successfully track object in complex conditions and its performances are better than MeanShift algorithm.更多还原