关键词: 轨迹完整提取, 车辆分类, 反馈修正

Abstract: The presented tracking system combines the vehicle’s type and trajectory feedback correction in order to obtain the complete trajectory in nighttime traffic surveillance. After filtering the restored headlight blobs in the distant, middle and close regions separately, the initial trajectories of headlights are obtained by area overlapping and trajectory feedback is then used for correction mismatching. Vehicles are classified according to the actual vehicle’s width which is calculated by homography matrix. Based on the temporal and spatial similarity, headlights are paired. The obtained trajectories of vehicles are finally corrected and optimized. The complete extraction of vehicles’ trajectories between the two adjacent monitoring points are achieved. Compared with the other similar algorithms, the number of frames that vehicles are tracked is increased by 500 frames, headlight matching rate is increased by 11.33% and vehicle tracking rate is increased by 10%.

Key words: complete trajectory extraction, vehicles classification, feedback correction