摘要
构建车牌字符识别系统,并对系统中BP网络反传学习速率进行改进,提高了识别率并降低学习时间;在特征提取上针对汉字综合采用非均匀网格特征和外围特征提取法,字母与数字采用均匀粗网格特征加笔划密度特征提取法,优化了系统的识别精度并提高了识别速度。采用BP算法增强了车牌识别的容错性、鲁棒性。
A license plate character recognition system is constructed, and the recognition rate is improved. Learning time is also decreased by improving the Back Propagation (BP) network. On the feature extraction, Chinese characters use non-uniform grid features supported by external feature extraction, letters and numbers use uniform coarse grid characteristics and strokes density feature extraction method. These feature extraction methods optimize the system and improve the accuracy of the identification of the speed. By using BP algorithm, fault-tolerant and robust of license plates recognition are enhanced.
出处
《电视技术》
北大核心
2008年第z1期140-142,共3页
Video Engineering
关键词
车牌识别
BP算法
特征提取
神经网络
字符识别
license plates recognition
BP algorithm
feature extraction
neural network
character recognition