Pedestrian abnormal event detection based on multi-feature fusion in traffic video

Pedestrian abnormal event detection is an active research area to improve traffic safety for intelligent transportation systems (ITS). This paper proposes an efficient method to automatically detect and track far-away pedestrians in traffic video to determine the abnormal behavior events. Firstly, pedestrian features are extracted by the multi-feature fusion method. Then, the similar features in current frame of all candidate objects are matched with the characteristic information of pedestrians in the previous frame which is considered as a template. Finally, pedestrian trajectory analysis algorithms are employed on the tracking trajectories and the motion information is attained, which can realize the early classification warning of pedestrian events. Experimental results on different traffic scenes in practice demonstrate that this method has good robustness in complex traffic. Moreover, the proposed method performs better compared with some other methods.