基于神经网络的汽车牌照自动识别

Automatic Recognition for Vehicle License Number Plate Based on Neural Network

介绍行驶车辆牌照自动识别系统基础上 ,对所获取的车辆正面车头图象 ,采用了多层次分割算法来定位车牌照 ,从图象背景中分离出牌照 ,再采用一种阴影掩膜技术对单个英文字母及阿拉伯数字进行编码 ,提取字符特征 ,并进而用BP网络进行识别。整个算法简单 ,识别率可达 96%。

The paper introduces an automatic recognition system of driving vehicle license number plate. From the front image of the vehicle we locate the place of the license plate by using multi-layer segment algorithm and separate the license plate from the background. Then we code the single English character and Arabic numbers by a shadow mask, pick up their characters, furthermore, recognize them by neural network. The whole algorithm is easy and the correct ratio of recognition is about ninety-six percent.