Research Of Pedestrian Detection And Tracking Method Of Driving Assistance System Based On Machine Vision

In recent years, the pedestrian detection and tracking technology based on machine vision applications in the field of intelligent vehicles more and more get the attention of people. It in a vehicle auxiliary driving system can effectively assist drivers to better understand the surrounding environment, to identify the pedestrians around, to have the possibility of pedestrian traffic accident early warning and improve the driving safety. So the pedestrian detection and tracking technology has very important significance.In this thesis, we focus on the machine vision auxiliary driving system of pedestrian detection and tracking technology. On the basis of the existing methods we put forward the corresponding improvement and innovation and ultimately realize the fast and exact testing results of pedestrians detection and tracking.The main innovations of this thesis are as follows:1. In terms of pedestrian detection, in this thesis, a two-dimensional Significant TS- LBP texture operator(Two- dimensional Significant Local Binary Pattern) is proposed, the operator can reflect the characteristics of the image texture characteristics, the significance and has strong anti-noise performance. First to extract the interested target area, then extracted the color and texture features of fusion of target area for target characteristics described, finally obtained a pedestrian classifier by Adaboost algorithm for pedestrians identification.2. In terms of pedestrian tracking, due to the light changes, the interference of similar backgrounds and other reasons, the traditional single color features MeanShift tracking algorithm will not be able to effectively track the target. This thesis proposes a partial significant SLBP texture operator(Significance of Local Binary Patterns), the MeanShift pedestrian tracking algorithm with color features and texture features fusion and through the adaptive weighting factor to adjust the convergence of the tracking target characteristics. The method is fast accurate and has high robustness, can carry on the effective tracking for pedestrians.