一种带有虚节点的HMM汉字识别后处理算法  

A New Algorithm for Contectual Postplocesing of Chinese Text by Using Virtual Nodes in HMM

作　　者：李宏东[1] 叶秀清[1] 顾伟康[1] 路浩如[1] X.S.Ma 

机构地区：[1]浙江大学 [2]Dept.of Appl.Mathematics,UniversityofVermont.USA

出　　处：《信号处理》1999年第3期254-259,共6页Journal of Signal Processing

摘　　要：研究一种基于ＨＭＭ的汉语文本上下文相关后处理新算法．通过引入虚节点，改进Ｖｉｔｅｒｂｉ动态规划算法，简化了搜索操作，提高了搜索效率。文中还给出了对手体汉字识别系统的后处理实验测试，其结果令人满意，表明本算法的有效性和实用性。So for it has been a common knowledge to take age of contectual postprocessing to enforce the ability of Chinese text processing, which can be used in the areas of recognition, speech recognition, and machine translation, etc. A new and efficient Algorithm for contoxtual postprocessing of Chinese text is developed and proposed in this paper that is based on the Hidden Markov Models (HMM). we modified the Viterbi Searching method by introducing Virtual Node concept. This modification allows us to building a normalized searching path (in term of path length), which results in a fast and efficient searching procesure. Experimental re sults on a handwritten Chinese character recognition system reach a correct rate above 10 percent which demonstrates the proacticable of our algorithm.

关 键 词：后处理 中文信息处理 汉字识别 HMM 算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]