Simple object coordination tracking based on background modeling

Abstract

In this paper, we propose a simple object tracking based on background modeling using histogram matching. Different from the existing block-based background modeling methods, most researches focus on a background subtraction. However, that includes a problem on visually object tracking and unreliable coordinate information. In this work, we implement background modeling and generate `background map' to reduce processing time for labeling. Then, `background map' is used for labeling algorithm to find a foreground that find object coordination information each frame. In addition, we can track moving object between previous frame and current frame using histogram matching. In the real-time processing, for a 640*480 resolution video, processing time is within 19ms using parallel studio.