基于贝叶斯原理的视觉多目标跟踪研究

Multi-Objects Vision Tracking Based on Bayesian Theorem

阐述了基于贝叶斯原理的粒子滤波技术。推导了粒子滤波的过程,将状态区域中包含的图像部分的直方图作为检测量与参考目标直方图比较,并在滤波过程中更新,得出目标最大后验估计。使用窗口平均法确定目标的最终坐标。在实验中,对单一及多个目标进行跟踪,结果证明,本文方法效果良好。

Propose particle filter technology based on Bayesian theorem. List the process of particle filter. The histogram in the image region of the state is regarded as measurement object that is eompared with the referenee image histogram. It is updated in the filter processing to obtain the optimized posteriori probabilities. The robust mean teehnique is applied to aseertain the objects＇ positions. In the experiments, both single and multi objects are traeked and the results show that the method in the paper is effieient.