Abstract: We presented a pedestrian detection method based on part detector and substructure assemble which can be used for driving assistant system or video surveillance system. Firstly we used the head classifier to search the whole image in order to get the ROI(Region of Interesting).Then in each ROI five part detectors including head, torso, leg,left arm and right arm were used individually. Finally we verified the detection result through part detector assemble method based on substructure. Experiments on different database show that our method has high performance in detecting pedestrians with numerous poses and partial occlusion in cluttered background.

Key words: Video images, Pedestrian detection, Part detector, Substructure assemble