基于结构特征和灰度特征的车牌字符识别方法

提出了基于结构特征和灰度像素特征的车牌字符级联识别方法。为提高车牌字符识别性能,分别在车牌二值小字符图像上提取结构特征和直接利用PCA降维后的车牌二值小字符图像的像素特征作为输入,用支持向量机(SVM)将其映射至高维空间分别进行分类,取两者中置信度高的结果作为预分类结果。当分类结果为8、B这类易混的字符时,取对应的灰度小字符图像像素值作原始特征,用PCA降维后再次用SVM进行分类,分类结果作为最后的二次分类结果。若为0、D时,则再次利用结构特征分类器做最后分类。该算法被用于台湾地区车牌的字符识别系统中,实验表明它能有效提高易混字符的识别正确率,具有很高的识别性能,应用前景广泛。

License Plate Character Recognition Based on Structural Features and Grayscale Pixel Features Algorithm

This paper proposed a new method based on the structural features and gray-scale features.To get a higher performance,the structural features and binary pixels features were extracted from the binary images respectively.These features were mapped to the high dimensional space through SVM to get the category.Gray pixels features were needed as an input of SVM while the category was8 or B.If the category was 0 or D,we used the 0D-classifier to classify it again.The algorithm was used to classify the license p...