基于条件随机场的低图像质量车牌字符分割

SEGMENTING CHARACTERS ON LICENSE PLATE WITH LOW IMAGE QUALITY BASED ON CONDITIONAL RANDOM FIELDS

提出一种基于条件随机场的车牌字符分割算法,能够对光照不均、相机拍摄角度造成的低图像质量的车牌图像,特别是日益增多的车牌边框与字符相连接车牌图像进行有效的字符分割。算法首先进行车牌图像校正,然后利用标注车牌数据进行模型学习,对车牌图像像素列进行分类识别,最后组合成车牌字符分割结果。理论分析与实验结果验证了算法的有效性。

This paper proposes a conditional random fields-based character segmentation method, it can make effective characters segmentation for the license plate image with low image quality caused by uneven light and shooting angle of camera, etc., in particular, for those increasingly growing license plates images whose plate characters are connected with plate borders. First, the correction of license plate image is conducted, and then the labelled plate data are used to train the model for the classification and identification of the pixel columns in license plate image, finally the segmentation results of the license plate characters are combined. Theoretical analysis and experimental results all verify the effectiveness of the algorithm.