Study On License Plate Recognition Technology Based On Support Vector Machine

With the continuous developing of the urban traffic, the license plate recognition system shows more essentialities and requirement, especially in the accurate rate and speed of the recognitions. The traditional technology of reconnaissance is hard to meet the necessity of modern traffic control.The license plate recognition system mainly includes three portions: license plate tilt correction, character segmentation and character recognition. The paper adopt Hough transform as the method of license plate tilt correction, and the character segmentation is achieved by the concept of the similarity between a character string and license plate. Also implemented is the license plate character recognition by using support vector machine(SVM) based on kernel clustering algorithm and overlap coefficient, the cumulate misclassification and the amount of the binary classifiers is reduced. Based on the research on the composing conditions and the characteristics of kernel functions, the arithmetic in the license plate recognition is improved by using the compound kernel function. Some conclusions have been made by a mass of experimentations: the new license plate recognition system obtains higher accurate rate and recognition speed, and the performance of the license plate recognition system is obviously advanced.