针对传统粒子滤波算法中容易发生的退化现象和粒子贫化问题, 提出多区域采样目标跟踪方法。该算法将目标模板用多个重叠子区域划分, 每个子区域对应一个采样窗口, 根据采样子区域置信度能有效估计出跟踪目标的真实状态, 子区域的互补性和阶段唯一性能很好地保证采样粒子有效性和状态空间质量, 从而提高目标跟踪的精确度。实验结果表明, 本文所提出算法能有效缓解目标跟踪中的粒子退化和贫化问题, 提高粒子利用率, 并且对目标形变、光照变化和部分遮挡等复杂情况具有较好的跟踪性能。

A particle filter for object tracking based on multi-region sampling is proposed to solve the problems of degeneracy phenomenon and particle impoverishment introduced by traditional particle filter algorithm. The proposed method uses some overlapping sub-regions to divide the target model, and each sub-region corresponds to a sampling windows. The true state of target can be estimated by the confidence of each sub-region. The complementary and stage uniqueness of sub-region can guarantee the validity of particles and the quality of state-space. Thereby, the accuracy of object tracking is improved. Experimental results show that the proposed method relieves effectively the sample degradation and poverty problems, improves the efficiency of particles, and is robust to pose, illumination and partial occlusion in the complex background.