视频监控中多运动目标的检测与跟踪

摘要

According to recursive-based pixel-labelled algorithm of connected region in binary images, we integrate the technologies of image pre-treatment, and improve the pre-treated images. Meanwhile, the motion object is better positioned in the image of real-time video surveillance according to its geometry feature, so as to realise detecting and tracking multiple motion objects. Experimental results show that the presented method is efficient and reliable, and can also satisfy the demands of the real-time video surveillance systems.%在基于递归方法的二值图像连通区域像素标记算法基础上,融合图像预处理技术,并对预处理后的图像进行改进,同时根据运动目标的几何特征,在实时监控视频图像中更好地定位运动目标,从而实现多运动目标的检测与跟踪.实验结果表明该方法可靠高效,也可以满足实时视频监控系统的需要.