【摘要】 根据车牌实际特征即车牌汉字为50个汉字中的一个且字体一致的特点,对汉字进行了统计特征分析,提出了适用于神经网络字符识别系统的汉字特征提取和复合特征分类方法。识别准确率和效率相对于其他特征提取方法有很大提高。更多还原

【Abstract】 The statistical features of Chinese characters are analysed in terms of the actual characteristics that the total number of Chi- nese characters for car’s license is 50 and their typeface is consistent.The method of feature extraction of Chinese characters and compound characteristic classification is put forward,which is suitable for the character recognition system by neutral net- work.The recognition accuracy and efficiency of this method are improved compared with other feature extraction methods.更多还原