New method of vision based vehicle detection and tracking in complicated background

Vehicle detection and tracking is key technology of vision based traffic monitoring systems. In this paper, a new scheme of vehicle detection and tracking in complicated background is proposed. Firstly, motion segmentation is performed combining background subtraction with motion edge detection. Then vehicles are identified in term of some proposed rules, which determine whether adjacent regions are with respect to the same vehicle. In addition, some characteristics of vehicles are recorded for vehicle tracking. Under the concepts of "confirmed vehicle" and "temporary vehicle", the proposed method succeeds in recovering momentarily missed vehicles as well as in distinguishing overlapping vehicles.