Fast moving object counting from video

At present, intelligent video surveillance is widely used in traffic scenes, such as streets, supermarkets, parking lots etc. The focus of this paper is how to quickly count the moving objects in the video image. Based on the mixture Gaussian modeling background subtraction algorithm, the foreground and background of the motion are separated, and the binary mask of foregrounds is obtained. Then, the breadth-searching algorithm is applied to count foreground moving objects in the binary foreground image. Experiments show that the processing speed of the proposed method is 40 fps and the average counting accurate rate is 84.27%.