Multi-target tracking with mobile camera using convex combination detection and nearest distance decision

Moving targets detection is one of the most important research subjects in video processing. This paper proposes a convex combination detection model, which uses the horizontal deviation and vertical deviation to magnify the difference between the target and background based on the grey scale model. We analyses the theoretical feasibility of the model, and then extract the targets from background using mask technology. We also propose a spot tracking approach based on our detection model. At first, we set up a fixed-size tracking window and calculate the integral image to locate the center of the target. Secondly, we use extreme value replacement method to decide the center of all the targets. Finally, we develop a multi-target tracking test platform which is able to implement detection and tracking of moving targets with various kinds of videos by setting different model parameters.