| 基于分层时序模型的步态识别算法 |
| |
| 引用本文： | 陈昌红,梁继民,朱秀昌. 基于分层时序模型的步态识别算法[J]. 南京邮电大学学报(自然科学版), 2011, 31(6): 1-6 |
| |
| 作者姓名： | 陈昌红 梁继民 朱秀昌 |
| |
| 作者单位： | 1. 南京邮电大学通信与信息工程学院,江苏南京,210003
2. 西安电子科技大学生命科学与技术学院,陕西西安,710071 |
| |
| 基金项目： | 国家自然科学基金(61172118,61001152);江苏省自然科学基金(BK2010523);南京邮电大学校科研基金(NY210073);江苏省高校自然科学基金(11KJB510012)资助项目 |
| |
| 摘 要： | 以隐马尔可夫模型和动态纹理模型为代表的动态贝叶斯网是描述步态序列的重要方法,但都存在一些不足之处.提出了一种新的动态贝叶斯网——分层时序模型,该方法采用分段线性逼近非线性和用各段的动态纹理模型作为隐状态,将隐马尔可夫模型和动态纹理模型做了结合,充分发挥了其优势.该方法在CMU Mobo步态数据库和CASIA步态数据库B上做了评估,结果充分显示了分层时序模型的高性能.
|
| 关 键 词： | 步态识别 分层时序模型 隐马尔科夫模型 动态纹理模型 |
Gait Recognition Based on Layered Time Series Model |
| |
| CHEN Chang-hong , LIANG Ji-min , ZHU Xiu-chang. Gait Recognition Based on Layered Time Series Model[J]. JJournal of Nanjing University of Posts and Telecommunications, 2011, 31(6): 1-6 |
| |
| Authors: | CHEN Chang-hong LIANG Ji-min ZHU Xiu-chang |
| |
| Affiliation: | 1.College of Telecommunication & Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China2.School of Life Sciences and Technology,Xidian University,Xi’an 710071,China |
| |
| Abstract: | Dynamic Bayesian networks(DBN),such as Hidden Markov model(HMM) and dynamic texture model(DT),are significant methods to model dynamic image sequence.However,they have some shortcomings.In this paper,we propose a novel layered time series model(LTSM),which takes piece-wise linearization to approach nonlinearity and employs DT of each piece as the hidden state.The new model combines DT and HMM,to make full use of their advantages.We evaluate the method on the CASIA gait database(dataset B) and the results show its good performance. |
| |
| Keywords: | gait recognition layered time series model hidden Markov model dynamic texture model |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
