Research On Vehicle Identification Technology In Surveillance Video

With the development of society, the city population is getting more and more intensive, the attendant traffic problem becomes more and more serious, intelligent transportation system is beening introduced to the traffic management, and vehicle detection technology and vehicle recognition technology are two key technology of the intelligent transportation system. The research studies vehicle identification technology of ITS, based on digital image processing, design a vehicle recognition prototype system. Vehicle recognition system consists of vehicle detection and extraction, image preprocessing, feature extraction, feature classification. Mainly completes the work:1. Vehicle detection and extraction. Extract the reference frame image from the video of the camera, compare a pixel in the current frame and the reference frame image, when the number of largely different pixels exceeds the threshold, the target vehicle is detected, then the background difference method for extraction of the target vehicle.2. Image preprocessing. In this paper, transform the color image into gray image with the weighted average method, and then use median filter for image denoising, then the gray stretch, histogram equalization method to enhance the image contrast, conducive to the subsequent feature extraction and comparison.3. Feature extraction. This paper focuses on the research of geometric features and texture features of the image, through the extraction of the vehicle width, high feature, according to the characteristic of wide high ratio to classify the different models of buses, vans and cars; by generating standard car face image gray level co-occurrence matrix, extracted five texture features from the gray level co-occurrence matrix, based on texture features to classify the different brands of Volkswagen, Toyota, Chevrolet, and Nissan.4. Feature classification, the paper extracts ratio and texture features of the two training sample library image, and compute the average ratio of every type of vehicle and the average texture feature vector, then use the minimum distance classifier, according to a ratio or the texture feature vector of the vehicle image to classify vehicle.This paper use VS2010, design and implement two sets of vehicle recognition system based on OpenCV, carry out the experiment of the acquisition of vehicle video image, the experimental results show that the correctness of the taken recognition algorithm, show that the system has good application value.