|
|||||
|
|
| 一种基于BP—HMM的字符识别方法 | |
| 引用本文： | 谢光军,秦江敏,杨江平.一种基于BP—HMM的字符识别方法[J].计算机工程与应用,2002,38(5):94-96,163. |
| 作者姓名： | 谢光军 秦江敏 杨江平 |
| 作者单位： | 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 |
| 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。 |
| 关 键 词： | BP—HMM K—L变换 贝叶斯决策 前(后)向评估算法 模糊度 |
| 文章编号： | 1002-8331-(2002)05-0094-03 |
The Method Based on BP-HMM for Text Recognition |
|
| Xie Guangjun,Qin Jiangmin,Yang Jiangping.The Method Based on BP-HMM for Text Recognition[J].Computer Engineering and Applications,2002,38(5):94-96,163. | |
| Authors: | Xie Guangjun Qin Jiangmin Yang Jiangping |
| Affiliation: | Xie Guangjun 1 Qin Jiangmin 2 Yang Jiangping 31 |
| Abstract: | 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. |
| Keywords: | Back Propagation-Hidden Markov Model(BP-HMM) K-L Transform Bayesian Decision-making Forward(Backward)Estimation Arithmetic Degree of Illegibility |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ | |