摘要： 为了提高视觉跟踪方法在物体外观发生变化时的性能，提出一种基于增量式子空间学习的视觉跟踪系统。该系统利用基于增量式主成分分析的粒子滤波方法增量式地学习一个表示跟踪结果的低维特征空间，以反映目标物体的外观变化。实验结果表明，当目标物体在复杂环境中承受姿态和光照变化时，该视觉跟踪系统具有更好的性能。

关键词: 增量式子空间学习, 增量式主成分分析, 粒子滤波

Abstract: In order to enhance the performance of visual tracking methods with object appearance variation, this paper proposes an visual tracking system based on incremental subspace learning. By using particle filter method based on Incremental Principal Component Analysis(IPCA), this system incrementally learns a low dimensional eigenspace representation of the tracking results to reflect appearance variation of the target object. Experiments demonstrate that the system has better performance when the target objects undergo large pose and illumination changes in some complex environments.

Key words: incremental subspace learning, Incremental Principal Component Analysis(IPCA), particle filtering

中图分类号: 

• TP391