Research On Key Technologies In Traffic Image Scenario Understanding

With the flourish of intelligent traffic monitoring technology, it brings the number of traffic surveillance images and videos growing rapidly. It is time-consuming and labor-intensive to analysis all the videos manually. Intelligently fast understanding and managing traffic images and videos are facing a great challenge. Traffic scene oriented image understanding is the ground for traffic image/video intelligently retrieval and management, and it is one of the key technologies to be solved in realizing intelligent monitoring. So the research of traffic scene oriented image understanding has theoretical and practical value. The main purpose of this paper is to explore the key technologies in the scene understanding of traffic scene images. The main research contents of this thesis are as follows:In view of the problem of redundant information in traffic scene image, this paper uses the detection method based on Local feature which makes full use of human eye perception mechanism to detect significant target.After the saliency detection, a new method of vehicle detection based on color histogram is proposed in this thesis. Firstly, the color space of candidate regions is converted from RGB space to HSV space. Based on the analysis of the color distribution of the vehicle, the H value probability distribution map of the candidate region can be identified, and the results can be obtained. The algorithm is simple and effective and has high accuracy.In view of the significance of vehicle symmetry detection to traffic scene understanding, this paper presents a method of vehicle symmetry axis detection based on clustering analysis. Firstly, the AdaBoost classifier is used to obtain the approximate position of the vehicle, and then the feature points are extracted by Harris corner detection method. Then the K-means method is used to extract the feature points. The feature points are extracted by random sampling method, and the relationship between the feature points and the symmetry axis is analyzed. In the end, the exact position of the axis of symmetry is found by establishing the kernel function of the symmetric axis parameters.Experimental results show that the proposed method can effectively detect the vehicle in the traffic scene, and can accurately locate the symmetrical axis of the vehicle.