基于字型特征的手写体汉字多分类识别的研究  

Study of handwritten Chinese characters multi-classification recognition based on structure features of characters

作　　者：王建平[1] 王晓雪[1] 

机构地区：[1]合肥工业大学电气与自动化工程学院,合肥230009

出　　处：《计算机应用》2007年第12期3084-3088,共5页journal of Computer Applications

摘　　要：针对汉字特点,提出一种基于汉字结构度和繁简度二类模态判别的多模式识别法。给出了汉字字型结构度类型的字型编码,以及汉字字型结构分解算法;对分解后的部件进行繁简度判断,依据各部件繁简度模态选择合适的特征提取算法,实现手写体汉字字型分解的多模式识别方法融合;对相似字采用两级分类的识别法,从而提高汉字的识别率和正确率。仿真实验验证了方法的有效性。A multimode recognizing method for two-type mode discrimination was presented based on the structure and complexity degree of Chinese characters. The type code and decomposing arithmetic of Chinese character structure were given. Character components were judged by complexity. Moreover, based on component complexity apiece, appropriate character-pick-up methods were selected to implement handwritten Chinese-character muhimode recognizing methods fusion on character structure decomposer. Similar characters were judged by two stages classification to improve the recognition efficiency. The method was proved in effect by emulation test.

关 键 词：汉字结构度 部件繁简度 多模式识别 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]