| Pedestrian detection has received more and more attention in the field of computer vision, pedestrian detection has become a hot research direction. It is used more and more in the transportation, security, statistics, and intelligent human-computer interaction and so on, and it will have wider application prospects. Under the current technical conditions, the machine vision technology is facing the complex external environment, the non-rigid of the identify target and complex trajectory, poses a great challenge to the machine vision recognition technology. In practice, video pedestrian detection algorithm which can be applied to the complex environment is not enough. Pedestrian detection algorithm robustness is to be improved, it put forward a new topic for researchers. This paper focuses on the Video pedestrian detection technology with non-mobile camera, From the view of digital video image background modeling, with the depth of image reconstruction based on laser Technology, analyses the foreground features, so as to complete the identification of pedestrians in video sequences.This paper mainly studied the design of video pedestrian detection technology based on embedded system, completed the design of the embedded platform TMS320DM642DSP hardware and driver, constructed a Pedestrian detection software development environment. According to the high demand of pedestrian detection algorithm based on embedded platform for real-time display and lean, designed the algorithm adapted to the embedded platform. Implemented a mixed Gaussian background modeling technology, and pedestrian characteristics recognition algorithm based on Freeman chain code on the DM642platform. In the part of depth of field assisted detection, this paper analyzes the current depth of field technology, and realizes the depth image reconstruction technology based on laser technology in the hardware platform. The video pedestrian detection assisted by depth image reconstruction technology played a good effect. After verification, the real-time and accuracy of video pedestrian detection technology in depth judgment based on DSP acquired preferably effect. |
