多特征与BP神经网络车牌识别系统研究

汽车牌照识别技术是智能交通管理系统中的关键技术;基于数字图像处理理论,将计算机视觉与模式识别技术相结合,对车辆牌照识别系统进行了研究。为了提高系统车牌识别能力,提出了一种综合颜色特征和投影特征相结合的车牌定位方法。字符分割采用了投影法;字符特征选取了互补性强的粗网格特征、投影特征以及外围轮廓特征;最后采用BP神经网络进行车牌字符识别。对车牌字符的识别分为汉字、字母及字母数字三类进行。实验表明,多种图像处理技术与模式识别技术有机结合能有效地提高系统的识别能力;本系统所采用的方法取得了较好的识别效果。

Research on Multiple Feature andBP Neural Network Licence Plate Recognition System

Vehicle license plate recognition technology is the key technology in the Intelligent traffic management system . The article which bases on the Digital image processing theory and combins the computer vision with pattern recognition technology does research into vehicle license plate recognition system . In order to improve the system vehicle license recognition ability,the article puts forward a method of License plate localization which combins comprehensive color feature with projection feature. Character segmentation uses the projection method . Character feature chooses the heavily complementary coarse mesh feature, projection feature and the peripheral outline feature. Finally ,we use BP neural network to recognize License plate character and the recognition consists of Chinese character, letter and letter number. The experiment shows that combining the variety of image processing technology with pattern recognition technology can improve the recognition ability of the system effectively.The method the system uses achieves preferable recognition results.