一种基于BP—HMM的字符识别方法  

The Method Based on BP-HMM for Text Recognition

作　　者：谢光军[1] 秦江敏[2] 杨江平[3] 

机构地区：[1]空军雷达学院研究生队,武汉430010 [2]空军雷达学院信息工程系,武汉430010 [3]空军雷达学院系统工程系,武汉430010

出　　处：《计算机工程与应用》2002年第5期94-96,163,共4页Computer Engineering and Applications

摘　　要：传统隐马尔柯夫模型广泛地应用在字符识别中,并具有较强的识别能力,但不能兼顾每个模型对其对应目标有很强的识别能力和模型之间差异性的最大化。该文提出的BP—隐马尔柯夫模型通过训练样本的不断训练,调整自身参数,解决了传统隐马尔柯夫模型不能解决的问题。计算机仿真结果表明:BP—隐马尔柯夫模型较传统的隐马尔柯夫模型有更强的抗干扰能力和更高的字符识别率。The traditional Hidden Markov Model is used widely in text recognition,and it has strong recognition ability.However it can't give attention to the strong recognition ability for the corresponding objects and the maximizing difference lain in different models.In this paper,the Back Propagation-Hidden Markov Model is put forward.We use training samples to train it repeatedly.At the same time ,this model adjusts its own parameters reiteratively.At the end,it solutes the problem perfectly that the traditional Hidden Markov Model can't do.The computer emulating results show that the Back Propagation-Hidden Markov Model has more powerful anti-jamming ability and higher text recognition rate than the traditional Hidden Markov Model.

关 键 词：BP-HMM K-L变换 贝叶斯决策 模糊度 字符识别 隐马尔柯夫模型 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]