摘要
为提高数字水表字符的识别率,提出一种基于数字图像处理技术与BP神经网络的水表自动识别系统;首先对采集的水表图像进行预处理以获得字符区域,然后进行二值化操作,利用连通区域提取算法分割出单个字符,最后利用训练好的BP神经网络对分割后的字符进行识别,并输出识别结果;实际运行表明自动识别系统正确率在95%以上,并且能够实时显示水表数据。
In order to improve the recognition rate of the digital water meter characters, a water meter automatic recognition system based on image processing and BP neural network is proposed in this paper. Firstly, characters area is located through pre--processing water meter image. Secondly, single character is segmented by connection component algorithm after binary thresholding. Finally, recognition re- sult is obtained by pre--trained BP neural network system. Experimental results show the accuracy of the proposed automatic recognition system more than 95%, and it can real time display meter data.
出处
《计算机测量与控制》
北大核心
2013年第5期1330-1332,共3页
Computer Measurement &Control
关键词
数字水表
字符识别
BP神经网络
连通区域提取
字符分割
digital water meter
character recognition
BP neural network
connection component extraction
character segment
