The Research And Application Of Vehicle Classification In ITS

A real-time vehicle classification system is designed and implemented in the thesis, which is a research focus and hard task in ITS. The system includes two main steps: training and classification. In the training step, first, using the time average image approach to get the background model, and then, in complex scene, using the background difference approach to detect and extract the outline of the moving vehicle, furthermore get the left, right and bottom edge of vehicle face according to the left, right and bottom edge of the outline, then, in the scene, select the same height threshold of the vehicle face and locate the vehicle face height. By processing the vehicle face image size normalized, the vehicle face image library is built. At last, extract the eigenvector of the vehicle face by using Eigenface method and construct the vehicle face feature library. In the classification step, first, vehicle face images are extracted by the above ways, then calculates the vehicle face eigenvector and compare with the eigenvector of the feature library using minimum distance method. The experimental platform is built with OpenCV and Visual C++, the vehicle face feature library is constructed by using 100 vehicles face in this thesis. In the experiment, adopting the size of 80*30 for vehicle face, the average time of one vehicle classification is about 1.88 ms in the system, when adopting the size of 120*50 and 322*131, the time is about 3.28 ms and 30.69 ms respectively. The experimental results show the feasibility and real-time performance of the system, when using the train library as the test library, classification rate was 100%.