汉字多属性认知特征提取研究及应用

摆脱传统汉字特征提取时基于二维平面象素点的统计和结构的分类方法.采用仿生模式识别思想,将汉字拆分为计算机算法可直接认知的基本笔划的组合,并对笔划进行数量、排序、位置,游程的属性编码,提出汉字多属性认知特征概念.方法很大程度上排除了传统汉字整体特征提取中信息的混淆和冗余,提高了字符识别正确性,实时性.用于车牌识别达到很好的实用指标。

Research on character multi-dim cognizing feature extraction and application

A method is presented to disassemble the character into several parts,which can be recognized by the computer topologically based on the high-frequency wavelet coefficients vector,disregarding the traditional extraction method using the statistical or structural feature based on the individual pixel in the 2-dim plane of character.Moreover,the concept of multi-dim cognizing feature is put forward by encoding the character,according to the location and run-length information.The information confusion and redundancy can be largely eliminated and the improvement of the preciseness,speediness and utilities is achieved when vehicle license plate is to be recognized.