摘要
文中提出了一种新的基于MQDF的车牌字符识别算法,该算法在QDF的基础上进行K-L变换,并且用常量代替小的特征值改善计算速度和分类的正确率。该方法基于统计模型和中心极限定理,便于设计和实现,广泛应用于手写体识别,具有很好的鲁棒性和较高的识别准确率。用2142幅白天、晚上的蓝牌、黄牌车牌图像做实验,实验结果表明,对于数字、字母、汉字字符,平均识别率达到98%以上,具有较好的应用前景。
This paper proposed a new license plate character recognition algorithm based on MQDF, which includes K-L transformation based on the QDF, and replaces the small characteristic value with constant to improve the calculation speed and classification accuracy. The method based on the statistical model and central limit theorem, which is easy to design, implementation, and widely used in handwriting recognition, has the very good robustness and higher identification accuracy. With the experiment of the license plate image of 2142 of the day, evening blue yellow card, the results show that, average recognition rate reaches above 98% for digit, letter, Chinese characters, which has a goodapplied foreground.
出处
《信息技术》
2013年第7期121-123,126,共4页
Information Technology
