基于Kalman算法改进的Camshift运动目标跟踪算法

Improved Camshift Target Tracking Algorithm Based on Kalman Algorithm

Camshift算法能够自适应更新跟踪框的大小、方向,在对特定颜色的目标跟踪时,能够取得很好的跟踪效果,但当目标受到背景因素干扰或者有局部的、不完全遮挡时,算法容易陷入局部最大值,导致丢失目标情况的出现最终跟踪失败,文章结合了kalman和Camshift算法,对目标有部分物体、环境遮挡等情况下的目标跟踪做出了改进,kalman做预测矫正得得到可能目标位置,并以此为中心确定Camshift算法的搜索区域,通过巴氏(Bhattacharyya)系数判别是否有遮挡情况出现并及时给予修正,在一定情况下对Camshift跟踪算法做出了改进。

The Camshift algorithm can adaptively update the size and direction of the tracking frame, and can achieve good tracking effect when tracking the target of a specific color, but the algorithm is easy to fall when the target is interfered by the background factor or has partial or incomplete occlusion. The local maximum value leads to the loss of the target situation and the final tracking failure. This paper combines the kalman and Camshift algorithms to improve the target tracking in the case of partial objects and environmental occlusion. Kalman makes predictions to obtain possible target positions. Based on this, the search area of the Camshift algorithm is determined. The Bhattacharyya coefficient is used to determine whether there is an occlusion situation and correct it in time. Under certain circumstances, the Camshift tracking algorithm is improved.