基于复合特征BP神经网络算法的车牌识别技术研究

Study on Vehicle License Plate Recognition Based on Hybrid Feature BP Neural Network Algorithm

传统的模板匹配算法对车牌图像的伸缩、倾斜及背景干扰比较敏感,识别效果不理想。提出将对字符进行归一化处理后所提取的点特征与重心特征进行复合,作为BP神经网络算法的输入特征,大大提高了识别率与识别速度。

The traditional template matching algorithm is sensitive to the stretching,tilt and background interference of vehicle license plate images,but the recognition result is not satisfactory.In this paper,through normalization processing of characteristics,the point feature and barycenter feature extracted are composted,as the input feature of BP neural network algorithm.The method greatly improved the recognition rate and recognition speed.