摘要
基于BP神经网络的车牌字符识别方法,是以BP神经网络为工具,使用图像的灰度特征及四角特征作为输入,实现对于字符的准确识别。相较于前人的特征提取方法,这种方法更适用于相似字符的识别。同时,使用Matlab提供的并行方法可以明显提升整幅车牌图像的识别效率。实验表明,对于相似字符的识别精度和整张车牌的识别速率方面,这种方法都具有更好的效果。
License plate character recognition method based on BP Neural Network is a way that uses the gray feature and the feature of four corners as input to accurately identify character. Compared with previous research, it is more effectively in similar character recognition. Meanwhile, This article used the parallel identification method proposed by Matlab to enhance the recognition’s rate of plate image. According to the data of experiment, the proposed method had better effect in the recognition accuracy and the rate of recognition.
作者
鹿琛
王姗珊
LU Chen;WANG Shan-shan(College of Computer and Information Technology/Shanxi University, Taiyuan 030006, China;Tongsheng Experimental High School , Linfen 041000, China)
出处
《山东农业大学学报（自然科学版）》
CSCD
2017年第1期113-116,共4页
Journal of Shandong Agricultural University：Natural Science Edition
关键词
字符识别
特征提取
并行识别
BP神经网络
Character recognition
feature extraction
parallel operation
BP neural network
