In the intelligent transportation system, the license information can be automatically recognized by the computer and the vehicle can be tracked. Red light running, illegal change of lanes, vehicle retrograde, and other illegal driving events are reasonably recorded. This is undoubtedly an effective help for the traffic police to relieve the huge work pressure. However, in China, a considerable number of vehicle tracking methods have certain limitations in resisting complex external environmental influences. The external environmental factors include but not limited to variable factors such as camera movement, jitter, and severe rain and snow. These factors cannot be controlled well, so the tracking accuracy is greatly reduced. In regard to this, this paper proposes an optimization method for moving vehicle tracking based on SPF. First, according to the size of the overlapping area of the motion area between the two images, the researcher can construct and simplify the vertex adjacency matrix that reflects the characteristics of the undirected bipartite graph. Then according to the corresponding relationship between the vertex adjacency matrix and the regional behavior and vehicle behavior, the researcher completes the regional behavior analysis and vehicle behavior analysis. On this basis, a particle filter vehicle tracking algorithm based on segmentation compensation is introduced, and the vector sum of the tracked segmentation area is used as the final position of the target vehicle. In this way, as many scattered particles fall on the target area as possible, which will greatly improve the efficiency of particle utilization, enhance tracking accuracy, and avoid the problem of tracking failure caused by too fast vehicle movement. Through experimental simulation, it can be seen that the method proposed in this paper can greatly enhance the vehicle tracking ability when tracking vehicles in “complex environments.”

With the development of the world economy and the acceleration of the urbanization process, the automobile has brought great convenience to people’s life and production activities and has become an essential means of transportation. Intelligent vehicles have the significance of reducing traffic accidents and improving transportation capacity and broad market prospects and can lead the development of the automotive industry in the future. Therefore, they have been widely concerned. In the existing intelligent vehicle system, lidar has become the leading role due to its excellent speed and accuracy and is an indispensable part of the realization of high-precision positioning. However, to some extent, the price is the main factor that hinders its marketization. Compared with the lidar sensor, the vision sensor has the advantages of fast sampling rate, light weight, low energy consumption, and low price; so, many domestic and foreign research institutions have listed it as the focus of research. However, the current visual-based intelligent vehicle environment perception technology is still prone to be affected by factors such as illumination, climate, and road type, resulting in the lack of accuracy and real-time performance of the algorithm. In this paper, the environment perception of intelligent vehicles is taken as the research object, and the problems existing in the existing road recognition and obstacle detection algorithms are deeply studied. Firstly, due to the complexity of texture feature extraction and voting calculation process of existing detection methods, and the influence of local strong texture feature interference inconsistent with road direction, a road image vanishing point detection algorithm based on combined 4-direction Gabor filter and particle filter technology was proposed. Then, aiming at the problem that the existing road image segmentation methods based on vanishing point constraint are too dependent on the edge features of road, which leads to oversegmentation easily, a method is proposed to improve the segmentation accuracy of road image by integrating road texture, road surface, and nonroad surface color features. Finally, the application of 3D reconstruction of road scene and obstacle detection technology based on binocular vision and visual navigation algorithm in intelligent vehicle trajectory tracking control is studied. Results show that the visual navigation algorithm can guide the vehicle routes along the road without a barrier, and compared with Wang Ren and two kinds of algorithm, the results show that this control algorithm effectively solves the traditional sliding mode control that is chattering phenomenon, overcomes the model matching, and does not match the interference problems, if used in the intelligent vehicle systems, it can reduce the thermal loss of electronic components and wear of actuator parts and improve the tracking accuracy.