改进的车牌相似字符分级分类识别算法研究

Research on improved algorithm of license plate similar character recognition by hierarchical and classified method

针对单一使用模板匹配车牌字符识别算法识别准确率低的缺陷,综合考虑到模板匹配方法的优劣,以及车牌字符集自身的特征,通过研究模板匹配方法的思想并对相似字符容易误识别的特点进行分析,采用改进的基于模板匹配车牌相似字符识别方法,并结合分级分类的思想,有效地提高了识别率。对不同环境下1030张高速卡口摄像头抓拍的车牌图像进行测试,提出的车牌字符识别方法准确率达到了系统的要求,并具有良好的鲁棒性和实时性。

In order to overcome the condition, where the identification rate of license plate character recognition is low by using the template matching algorithm singly, a improved algorithm of license plate similar character recognition is presented, consid- ering the advantages and disadvantages of template matching method, and the characteristic of license plate set. Based on hierar- chical and classified idea, the template matching algorithm and the characteristic of similar character mistaken are researched. The identification rate is improved effectively. In the experiment on identifying license plate number, 1030 images in different en- vironment are used. The result of the experiment shows that the algorithm presented ensures the requirements and has a better robustness and real-time, so it satisfies the needs of the project.