| 基于一种改进禁忌搜索算法优化离散隐马尔可夫模型 |
| |
| 引用本文： | 刘江华,陈佳品,程君实. 基于一种改进禁忌搜索算法优化离散隐马尔可夫模型[J]. 计算机工程与应用, 2003, 39(20): 92-94 |
| |
| 作者姓名： | 刘江华 陈佳品 程君实 |
| |
| 作者单位： | 上海交通大学信息存储研究中心,上海,200030 |
| |
| 摘 要： | 隐马尔可夫模型(HMM,HiddenMarkovModel)是语音识别和手势识别中广泛使用的统计模式识别方法。文章提出了一种改进的禁忌搜索(ITS,ImprovedTabuSearch)优化HMM的参数。传统的TabuSearch(TS)与局部搜索算法(极大似然法)交替进行,从而加快了算法的收敛速度,并得到优化解。分别用TS及ITS训练隐马尔可夫模型进行动态手势识别。结果表明ITS可获得更高的识别率,且能达到全局优化。
|
| 关 键 词： | 改进的禁忌搜索 隐马尔可夫模型 动态手势识别 |
| 文章编号： | 1002-8331-(2003)20-0092-03 |
| 修稿时间： | 2002-06-01 |
Optimization of Discrete Hidden Markov Model Based on an Improved Tabu Search Algorithm |
| |
| Liu Jianghua Chen Jiapin Cheng Junshi. Optimization of Discrete Hidden Markov Model Based on an Improved Tabu Search Algorithm[J]. Computer Engineering and Applications, 2003, 39(20): 92-94 |
| |
| Authors: | Liu Jianghua Chen Jiapin Cheng Junshi |
| |
| Abstract: | Hidden Markov model(HMM)is a statistical pattern recognition method that is extensively used in speech and gesture recognition.In this paper,an improved tabu search algorithm(ITS)is used to optimize the parameters of HMM.Traditional tabu search algorithm(TS)is combined with local search algorithm,maximum likelihood(ML )algorithm,which speeds the convergence of the algorithm and can get better solutions.TS and ITS are applied to training HMM for dynamic gesture recognition.The results show that ITS can get higher recognition rate,and find the optimal model parameters. |
| |
| Keywords: | Improved tabu search Hidden Markov model Dynamic gesture recognition |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
