一种基于相似度判据的K近邻分类器的车牌字符识别方法

针对K近邻方法分类准确率较高的特性,提出了一种基于相似度判据的K近邻分类器车牌字符识别方法.通过大量实验,选取字符的网格特征和轮廓特征作为分类依据,用特征融合方法将两种特征合并,实现特征的串行融合.根据相似度判据作K近邻分类,实现了一个车牌字符识别系统.实验表明,这种方法具有良好的识别效果,鲁棒性强,具有较大的实用价值.

An Algorithm Based on Similarity and KNN for Licence Plate Recognition

As the K nearest neighbor method is of high classification accuracy, proposed a algorithm based on similarity and KNN for licence plate recognition. Choose profile feature and griddling feature as the classification features of the licence plate recognition. Then fuse the features. According to above feature fusion, An Algorithm based on similarity and KNN is used to construct a licence plate recognition system. Using the algorithm, a high recognition rate can be reached. Experimental results demonstrate the algorithm is feasible, robust and applicable.