一种复杂背景下运动目标检测与跟踪方法

Detecting and tracking method of moving target in complex environment

为了准确快速检测和跟踪运动目标,扩大视频监控系统中摄像机的搜索和跟踪范围,在目标检测部分提出了一种将双向差分相乘法与基于仿射运动模型的匹配差分法相结合的运动目标检测方法;在目标跟踪部分改进Camshift算法,利用目标边缘轮廓信息完成对运动目标的跟踪,最后在基于BF533的网络视觉跟踪监控系统上进行了实验。结果表明,所提方法能够对运动目标进行准确检测和大角度跟踪,在单目摄像机的视频监控系统中具有较高的实用价值。

In order to detect and track moving targets accurately and rapidly, and to expand searching and tracking scope of the camera, this paper presents a moving target detection approach, which combines two-way difference multiplication method and matched difference method based on affine motion model. In motion tracking of targets, the Camshift algorithm is improved, and the edge information of the outline of the campaign targets is used to track the moving objects. Finally the approach given in this paper is tested on an Internet vision tracking surveillance system based on BF533. Experiment results show that this approach can detect moving targets accurately and track them in a large bearing, and has high practical value in monocular video camera monitoring systems.