对车牌中的字符进行精确提取并识别是车牌识别系统面临的重要问题。在字符提取阶段, 利用车牌图像特有的信息特征, 提出以红色分量作为目标图像, 依次进行水平和垂直投影完成字符分割, 既可精确提取字符, 又可矫正字符形态。将基于结构相似度SSIM(structure similarity)的算法应用于车牌识别, 避免了前期对字符结构信息做大量的对比统计工作, 省略了特征提取的步骤, 降低了算法复杂度; 利用角点检测方法对相似字符进行细化和区分, 进一步完善了整体的识别性能。实验证明, 该算法有较好的识别率。

One of the main problems for license plate recognition system is to maintain an accurate view of extraction and recognition of the license plate. In the character extraction stage of image recognition, the red component of the license plate information is extracted as the target image depended on the unique characteristics of the license plate image, and the splitting characters by the horizontal and vertical projection will result in a more accurate extraction and correct character form. A new method based on the structure similarity recognition is introduced. This approach avoids a lot of statistical compare on character structure information in early process and reduces the computing complexity by omitting the step of feature extraction. A method of corner detection is given, which further perfects the integral recognition performance to distinct and refine similar characters. The experiment results show that the algorithm has the better recognition rate.