神经网络在车辆牌照字符识别中的应用

Neural Network Applied in Character Recognition of Vehicles' License Plates

在车辆牌照自动识别系统中 ,因自然因素或采样因素使得原本规则的印刷体字符产生畸变 ,给字符识别带来了很大困难。本文在特征抽取的基础上 ,采用 BP网络进行分类 ,并附加线性感知器来实现单字的有效识别。该方法算法简便 ,识别率高 ,可适用于多种高噪声环境中的印刷体字符识别。

In conventional automatic character recognition systems, normal printed characters tend to be distorted due to natural or sampling factors, which hinders their correct recognition. This paper applies BP Neural Network to classification based on feature extraction and with a linear perceptron for recognition of individual characters. It proves effective and efficient for printed character recognition in the high noise level environment.