PEDESTRIAN DETECTION BASED ON MODIFIED DYNAMIC BACKGROUND USING GAUSSIAN MIXTURE MODELS AND HOG-SVM DETECTION

In this paper, we propose a fast pedestrian detection method based on surveillance video clips under stationary cameras. Our purpose is to address the problem of low speed of pedestrian detection with an HOG-SVM detector. First, the background modeling using the mixture of Gaussians is used to extract the moving objects in the video. Then, three steps, i.e., shadow removing, eroding and dilating and border expanding are performed to make further alterations to the extracted foreground. At the same time, our experiments based on the INRIA dataset calculate the histogram of oriented gradients feature of the whole pedestrians and classify them by a support vector machine. Experimental results indicate that the foreground extracted by our background modeling scheme can contain all the moving objects well through shadow removing and border expanding. So the proposed methods outperform the traditional HOG+SVM method in both recognition accuracy and processing speed.