Real Time Pedestrian Detection Algorithm by Mean Shift

Abstract

Conventional moving objects detection algorithm associated with visible image is often affected by the change of moving objects’ shapes, illumination conditions and is also influenced by complex backgrounds, shadow of moving objects, moving objects of self-occlusion or mutual-occlusion phenomenon. This paper presents a method of human detection by mean shift based on depth map. By analyzing and comprehensively applying segmentation method based on height information to extract moving target and remove the background information from depth map, the region of interest (ROI) with moving target should be found, then through mean shift method the goal of real-time objects (pedestrian) detection can be achieved eventually. In this paper, using the depth image pattern recognition is a good way to overcome the difficulties that visible light image pattern recognition often encounters. The depth image pixel value is only related to the distance from the surface of the object to the view window plane. Therefore, depth image has nothing to do with color space and does not suffer from the factors such as illumination, shadow effect. In addition, the mean shift targets detection method with high efficiency and fast speed features can solve the problems of low identification efficiency and poor real-time performance based on traditional pedestrian detection system to a certain extent. Our algorithm using mean shift method based on depth information has been tested on several image sequences and shown to achieve robust and real-time detection.