Study Of Vehicle Detection And Tracking Based On Vision

Real-time vehicle detection and tracking based on computer vision is an important research area in intelligent transportation systems and an important method to reduce the traffic accidents. It has promising prospect in driving assistance and accident warning field. In this thesis, upon the background of driving assistance on highway, we present a new framework for object detection and tracking on the basis of road area analysis. Based on this framework, we proposed vehicle detection and tracking algorithms.At the beginning of this thesis, we firstly introduce the development of intelligent vehicle oversea and domestic. At the same time, some research institutions that are related to the intelligent vehicles are listed to show their products of intelligent vehicles. Then some applications of computer vision are discussed in the orientation assistance navigation system, the self-adapted cruise control system, etc.This thesis discusses the technology of road recognition based on the Hough transformation. Comprehensively we give the analysis of varieties of image pre-processing technology, and through comparative analysis of the experiment, we choose suitable pre-processing algorithm for the subject; Final we give the analysis of the probabilistic Hough transform and its dominance in road edge detection, and give a test simulation. The experimental results show the robustness of this algorithm.This thesis proposes a vehicle detection proposal based on the extraction of shadow with a combination of vehicles' texture and symmetry of gray-scale. First of all, an adaptive threshold method is used in the segmentation of the shadow at the bottom of a vehicle. Then it builds areas that vehicle may exist in the candidate region; Secondly, in the scope of the road, the candidate region are filtered by texture characteristics; Finally, the gray symmetry of the vehicle' rear is used to confirm the existence of the vehicle.Based on the vehicle detection algorithm, this thesis concentrates on the algorithm for vehicle tracking. Camshift algorithm, respectively, with Kalman filter are applied to vehicle tracking. We point out their advantages and disadvantage throngh testing. Then a framework of vehicle tracking that combining Camshift with Kalman filter is proposed, which can be reparative for each other. Good tracking effect is shown by varieties of testing for which the algorithm proposed in this thesis .Combined to the study of the anti-collision system of vehicle vision hold by our lab, this thesis does the research of road recognition of vehicle ahead and obstacle detection and tracking algorithm; and conducts the emulation test of the vehicle detection and tracking ahead and rear of the main car on the highway using PETS2001, which can provide a good support to the devolopment of anti-collision and warning system next.