多特征融合的粒子滤波视频跟踪算法研究

Research on the Particle Filter Video Tracking Algorithm Based on Multi-Features Fusion

提出了基于颜色和边缘双重特征的目标跟踪算法,研究了基于HSV模型的空间加权的颜色直方图和边缘特征的直方图提取算法,针对多个特征如何融合问题,设计了一种距离度量对视觉特征的可靠性评价,解决了传统的粒子滤波目标跟踪方法易出现跟踪丢失和漂移等不稳定问题,提高了跟踪算法的目标表示能力,增强了算法的稳定性.

In this paper,an object tracking algorithm based on dual features of color and edge is proposed,The spatial weighted color histogram and edge feature histogram extraction algorithm based on HSV model are studied. Aiming at the problem of how to integrate multiple features,the reliability evaluation of distance measurement for visual features is designed,and it solves the instability problem of traditional particle filter,such as tracking loss and drift. The target representation ability of the tracking algorithm is improved,and the stability of the algorithm is enhanced.