基于SOM神经网络聚类的车牌号码倾斜校正算法

经过摄像机摄入的图像会发生倾斜,给车牌的准确识别带来了困难.针对此问题,利用SOM神经网络良好的聚类性能,在水平倾斜校正时,把车牌号码图像中的像素坐标聚成两类,拟合成一条直线,计算出该直线倾斜角,完成水平校正;按照以上同样方法进行垂直倾斜校正.实验结果表明,该方法能准确获取车牌号码的倾斜角,算法结构简单,抗干扰能力较强,符合汽车牌照图像的特点,具有较好的处理效果.

Method of correcting license plate tilt based on SOM

The camera images will happen intake tilt, which made it difficult to recognize the license. According to this problem, a new method is proposed for correcting license plate tilt based on SOM neural network. To the method, the image character coordinates are divided into two clusters to fit a line and the line slope is obtained. The rotated image is done shear transform and the final correction image is obtained. The experimental results show that, this method can get the tilt angle accurately and has strong anti-jamming and practical application. The method will be more suitable for the features of the license plate and with better dealing efficiency.