Research And Application Of Vehicle Recording System Using Vehicle Classification

A vehicle classification system is designed and implemented in the thesis. Andthe system includes two main steps: establishment of the sample library and vehicleclassification. In the establishment of the sample library step, first, doing the imagepre-processing, then using the time average image approach to get the backgroundmodel, and then, in complex scene, using the background difference approach to detectand extract the outline of the moving vehicle, furthermore get the left, right and bottomedge of vehicle face according to the left, right and bottom edge of the outline, then, inthe scene, select the same height threshold of the vehicle face and locate the vehicleface height. By processing the vehicle face image size normalized, the vehicle faceimage sample library is built. In the vehicle classification step, first, vehicle faceimages are extracted by the above ways, then calculates the vehicle face sift featuresand compare with the sift features of the sample library. When the matching valuessatisfy certain conditions the vehicles can be determined to be the same type of vehicle.The experimental platform is built with OpenCV and Visual C++, the vehicle facefeature library is constructed by using100vehicles face in this thesis. In theexperiment, adopting the size of80*30for vehicle face, the average time of onevehicle classification is about150ms in the system, when adopting the size of160*60and320*120, the time is about450ms and1400ms respectively. The experimentalresults show the feasibility performance of the system, and in practical applications canget high recognition rate.