基于 Surendra背景减除法和四帧差分法的目标检测算法

摘要

研究监控环境下的运动目标检测算法，提出一种新的运动目标检测算法。利用四帧差分算法将连续四帧两两差分得到运动区域，与Surendra背景减除法确定的运动目标相结合，采用动态阈值以适应光线突变，通过连通性检测和孔洞填充进行后处理，综合得到运动区域图像。实验结果表明，该算法有效地去除了噪声、空洞及双影等现象，具有较好的实时性和鲁棒性。既能精确地检测出运动目标，又能迅速响应实际场景的变化。%We carry out the study for the moving object detection algorithm in monitoring condition, and propose a new moving object detection algorithm.It uses four-frame difference algorithm to make every-two difference on continuous four-frame to get motion regions, and combines them with the moving object determined by Surendra background subtraction, and employs dynamic threshold in order to adapt to the sudden changes in lights, and conducts post-processing through connectivity detection and hole filling, thus the motion region image is obtained synthetically.Experimental results show that, the algorithm can effectively remove noise, cavity and double shadow, and has preferable real-time property and robustness.Apart from accurate detection of moving object, it can also rapidly respond to practical scene changes.