The Research On License Plate Character Recognition Based On Support Vector Machines

Recognition of license plate characters is the last step in the license plate recognition system, and it requests that recognizing single character rapidly and accurately. The result of characters recognition shows the success or failure of the whole license plate recognition system. However, most of the methods is under traditional statistic mode, which is under the condition of adequate samples and empirical risk minimization theory. It could get sufficient recognition results only if the number of samples runs to infinite. Unfortunately, in practical issues such as license plate character recognition, the number of samples is limited, and then ideal results can hardly be got on the basis of the existing methods.Small sample learning problem can be solved by SVM very well. The aim of learning is not only to get optimal values when samples tend to be infinite, but also to get the optimal solutions under current conditions of information. When samples are linear non-separable, the input vectors are mapped to a high dimensional feature space using the predefined nonlinear mapping, and then one optimal hyper-plane is constructed in the high dimensional feature space. It mainly depends on the kernel function that is used by SVM and some complicated computing can be avoided in the high dimensional feature space.This thesis shows the theoretical of SVM classification and discusses several SVM classification algorithms. Specially, SMO algorithm is studied deeply and improved.To the steps of the license character recognition, we conduct normalized processing with Centroid and bilinear interpolation to the binary images that have been noise filtered firstly. Sencond, we present the method that the character geometric transformation is done, and then integrate the rough grid and direction of line feature to extract features. The input vector parameters of SVM are mainly constructed form rough grid feature of characters. Finally, character is classified using MLP and unilateral binary decision tree. The experimental result shows that the proposed approach is feasible and effective not only to figures and uppercase, but also to Chinese characters.On the basis of analyzing Multi-classification SVM, aiming at the question of a small quantity of image classification, we present a new Multi-SVM classification method based on MLP and unilateral binary decision tree. Through training of MLP network, Lagrange multiplier vector and threshold value b in decision function, constant in kernel function, restricted value in v-SVM classification can be gotten. In the end, character images are classified step by step using unilateral binary decision tree. When classifying, one procedure of adjusting parameter is adopted to shorten the time of classification and advance the precision of classification.