Moving object detection based on running average background and temporal difference
Yi Zheng, Fan Liangzhong · 2010
In order to detect moving objects from video sequences with complex background, we propose an algorithm which is based on running average background modeling and temporal difference method. Firstly, we utilize the running average method to dynamically updating the background image. Through using background subtraction, we get a foreground image. Secondly, we use temporal difference method to get a difference image. By combining the foreground image with the difference image, the common information between them can be achieved. Finally, we eliminate the noise in the combined image by using the median filter, and then we can get the moving objects. Experimental results show that, comparing with traditional running average method, temporal difference method and Gaussian mixture background modeling method, our method can detect the moving objects from complex backgrounds more accurately with low computational complexity.