中文摘要：

提出一种适用于被动式光学人体运动捕捉散乱数据处理方法。该方法基于光学人体运动捕捉散乱数据的全局信息，提出基于模块分段线性模型的数据处理算法。利用模块分段线性模型归纳出不同模块的变化特征，从而确定各模块数据的匹配优先级及段内拟合函数，有效地对三维运动数据各模块进行全局性分层次预测和跟踪，并对噪声数据进行基于模块的去噪处理；对缺失运动数据提出基于分段Newton插值拟合算法，进行合理的补缺。该方法经优化后在处理过程中无须人工干预，并能满足实时性要求。

英文摘要：

This paper presented a scattered data processing method for passive optical human motion capture. This approach was based on the overall information of optical human motion capture scattered data. According to piecewise linear model of different modules, determined the matching priority of each module data and fitting function of every section. Furthermore, predicted and tracked 3D motion data overall point level. At the same time, removed noise data. For missing data on the movement, provided a fitting algorithm for the missing motion data. The computer simulations illustrate that the data processing is in real-time and need no manual works.