摘要
为有效解决手势跟踪研究中的高维问题,提出一种基于交互行为分析的自然人手跟踪和交互算法。探讨了基于手势的交互行为分析和建模,建立了操作者手势的四阶段行为预测模型,揭示了手势交互的一般规律或关键特征,并给出建立行为模型的具体方法和过程;然后,研究基于提出的行为模型的手势跟踪和交互算法。采用数学方法对分段预测模型进行描述,提出多通道采样和预测手势跟踪算法。通过与几种相关算法进行的对比实验表明,该算法可以有效地改善手势跟踪的速度和精度。
To solve the high dimensionality problem in hand tracking research,a hand tracking algorithm was proposed based on interactive behavioral analysis.The analysis and modeling of interactive behavior based on hand gesture was discussed,and four stages behavior predicting models of handles＇ gesture were built.The general rules or key features of gesture interaction was revealed,and the concrete method as well as process to build behavior model were given.And then,the proposed model＇s hand tracking and interactive algorithms were studied.The mathematical methods were used to describe the segmented prediction models,and a multi-channel sampling as well as predicting hand tracking algorithm were proposed.Comparing with several related algorithms,the hand tracking＇s speed and accuracy were improved by using proposed algorithm.
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2012年第1期31-39,共9页
Computer Integrated Manufacturing Systems
基金
国家自然科学基金资助项目(61173079
60973093
60873089)
山东省自然科学杰出青年基金资助项目(JQ200820)
教育部新世纪优秀人才计划资助项目(NCET-10-0863)
国家863计划资助项目(2009AA043506-2)
山东省自然科学基金重点资助项目(ZR2011FZ003)~~
关键词
运动人手跟踪
行为模型
手势交互
粒子滤波
moving hand tracking
behavior models
gesture interaction
particle filtering
