License plate recognition system based on morphology and LS-SVM

This paper studies on license plate recognition (LPR) system based on morphology and least squares support vector machines (LS-SVM). LPR is one of the most important research topics in intelligent transport field. Firstly, locate the license plate based on the improved Robert edge operator and morphology. Secondly, segment the characters according to the projection and real information of the license plate. Finally construct several classifiers using LS-SVM to implement the license plate character recognition. Experiment results show that the location and segmentation is accurate and the average recognition speed is about 19.4 ms/character. Moreover the accuracy for recognition is higher.