多帧背景差与双门限结合的运动目标检测方法

Moving Target Detection Method Based on Multi-frame Background Subtraction and Doublethreshold

为有效解决视频监控场景下运动目标快速、精确检测的问题,提出一种多帧背景差与双门限结合的运动目标检测方法.首先改进Surendra背景模型来获取干净的背景图像,根据灰度差分图像确定的两个门限值进行前景目标检测,低门限阈值用于检测出比较明显的前景目标(即粗检测),在粗检测的基础上利用高门限阈值以去除粗检测中存在的噪声目标与伪目标(即细检测),最终实现视频监控场景下运动目标的精确检测效果.针对车辆、行人等不同对象的监控场景下进行实验,验证了本文方法不仅能够有效地抑制噪声及伪目标的干扰,而且能够快速、准确地分割出前景目标.

To effectively solve the problem of fast and accurate detection of moving targets in video surveillance scene,moving target detection method based on multi-frame background subtraction and double-threshold is proposed. Firstly, Surendra background model algorithm is improved to get clean background image, then foreground object is carded out by the two threshold values of gray differ- ence image,low-threshold value is used for detecting the obvious target （ i. e. , rough detection ） ; on the basis of the rough detection high-threshold value is used for removing noise target and false target （ i. e. , fine detection ）. Finally, moving target is detected accu- rately in video surveillance scene. The experiment on the vehicle, pedestrian and other object shows that the method can not only sup- press the noise and interference of false target,but also can segment foreground target rapidly and accurately.