A New Method of License Plate Characters Recognition by Feature Extraction and BP Neural Networks

摘要

In this paper, the feature of rough grid is used to enhance our character recognition system, These features is applied to improved unitary character originality feature to BP neural network classifier to recognize the license plate character.To improve the recognition rate of the character recognition system of vehicle license plate. By analyzing the result of test, we also put forward a series of new measures including designing a careful neural network classifier to distill detail feature of a character which is analogical and promiscuous, found on Chinese characters stroke conglutination and character excursion phenomena, based on standard samples. For some baroque strokes compactness province characters, we also properly add representative stroke conglutination samples and representative excursion samples to Chinese character's network training samples. The study result shows that the new method can greatly improve the steady performance of character recognition system.