The Research On Vehicles' License Plate Recognition System Based On Support Vector Machines

Support Vector Machines(SVM)is a new method for pattern recognition lately. It operates on the principle of structure risk minimization which not only keeps the empirical risk minimal but also controls VC confidence of discriminant function, hence a better generalization ability is guaranteed. And Vehicles' License Plate Recognition(LPR) system is an important application of computer vision and pattern recognition in the intelligent transportation management, the improved accuracy of LPR is required. Combining them is taken as the start of this article.In the article, SVM is considered as the research method and is used to solve some difficult problems in LPR system. I has finished following work in this article:1. Locating the vehicles' license plate plays an important role in LPR system, a novel locating approach based on SVM is presented .2. Two binarization methods are presented in this paper, iterative selection and Minimum cross-entropy method.3. Combine the plate's projection and the prior knowledge of vehicle's plate, a practical characters segmentation is proposed in this paper.4. Construct Multi-classifier based on SVM, then extract the feature of character to train the classifier, finally, give the result of experiment comparing with other methods.5. Implement the above algorithms in Visual C++6.0 based on the theory research and develop a software of Vehicles' License Plate Recognition System.The research in this article shows that using SVM classifiers can efficiently improve the robustness and recognition rate of the system comparing with other methods.