基于视觉的动态手势识别  

Vision based dynamic hand gesture recognition

作　　者：孔晓明[1] 陈一民[2] 陈养彬[2] 殷烨[2] 

机构地区：[1]山东大学威海分校信息工程学院,山东威海264209 [2]上海大学计算机工程与科学学院,上海200072

出　　处：《计算机工程与设计》2005年第11期2934-2936,共3页Computer Engineering and Design

基　　金：上海市科委重点攻关基金项目(015115042)

摘　　要：采用隐马尔可夫模型(Hidden Markov Model)训练并识别手势样本,针对HMM的经典训练算法Baum-Welch算法的缺点,采用纠错算法对其修正,提高了识别率。将识别结果应用于“基于Internet远程机器人控制”项目,达到了对机器人控制的目的,优化了人机交互的接口。The Hidden Markov Model （HMM） was used and the rectify method was also introduced to correct the training of HMM, and a better result was gotten. Finally, the results of dynamic hand gesture recognition were introduced to the “Internet based remote Robot control”project, which realized the objective of controlling robot and optimized the Human Computer Interactive interface,

关 键 词：隐马尔可夫模型 模式识别 光流跟踪 手势分割 阈值 

分 类 号：O235[理学—运筹学与控制论]