| 改进AdaBoost对基于HMM的唇读系统识别率的提高 |
| |
| 引用本文： | 鹿佳,姚鸿勋.改进AdaBoost对基于HMM的唇读系统识别率的提高[J].哈尔滨商业大学学报(自然科学版),2005,21(5):604-608. |
| |
| 作者姓名： | 鹿佳 姚鸿勋 |
| |
| 作者单位： | 哈尔滨工业大学,计算机科学与技术学院,黑龙江,哈尔滨,150001;哈尔滨工业大学,计算机科学与技术学院,黑龙江,哈尔滨,150001 |
| |
| 基金项目： | 国家863计划(2001AA114160)
哈尔滨工业大学校基金(HIT 2002．72)． |
| |
| 摘 要： | 针对目前唇读系统多采用HMM的识别方法,提出了基于AdaBoost的唇读识别方法,有效地解决了样本空间的交叠问题,通过强化训练那些难以分类的样本,使得识别性能有所提高.该方法改进了迭代过程中权值的变化率,降低了样本权重更新速度;同时区分噪声样本,减小不合理弱分类器的权重,使得改进后的算法降低了噪声对强分类器的影响.
|
| 关 键 词： | 唇读 AdaBoost HMM |
| 文章编号： | 1672-0946(2005)05-0604-04 |
| 修稿时间： | 2005年7月10日 |
Improved AdaBoost method to increase recognition rate of traditional lip-reading system based on HMM method |
| |
| LU Jia,YAO Hon-gxun.Improved AdaBoost method to increase recognition rate of traditional lip-reading system based on HMM method[J].Journal of Harbin University of Commerce :Natural Sciences Edition,2005,21(5):604-608. |
| |
| Authors: | LU Jia YAO Hon-gxun |
| |
| Abstract: | For HMM models are usually applied of lip-reading systems presently, a lip-reading recognition method based on AdaBoost is proposed in this paper. This method solves the overlap problem of sample space, and improves the performance of recognition by strengthening training the samples that are difficult to be classified correctly. The method mentioned above, improves the change rate of the samples' weights in the iteration process, reduces the updating speed of samples' weights; and it can distinguish noisy samples, and reduces the weights of unreasonable weak classifiers, so that the improved algorithm suffers less influence of noisy data on strong classifier. |
| |
| Keywords: | lip-reading AdaBoost HMM |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
