摘要： 针对虚拟人行走引擎的实时生成问题，在运动捕获数据的基础上，提出将主成分分析(PCA)方法引入到虚拟人的行走引擎研究中，实现通用的实时行走引擎。根据特定速度值，在生成的PCA空间中进行插值和外推计算生成新的运动，通过时间弯曲方法使引擎由髋关节高度参数来驱动。实验结果证明，采用基于PCA方法设计的行走引擎，可以实时生成比较流畅的行走动画。

关键词: 虚拟人, 主成分分析方法, 插值, 外推, 时间弯曲, 运动捕获数据, 行走引擎

Abstract: In this paper, according to the real-time generation problem of walking engine of virtual human, based on the motion capture data, the Principal Component Analysis(PCA) method is introduced into the research on generic walking engine of virtual human, and realize real-time walking engine. Interpolation and extrapolation are achieved in the PCA spaces of each subject to generate new motion according to a speed value. A time warping method allows handling the human height parameter. Experimental result shows that the walking engine based on PCA method can produce smooth walking motions with real-time reactivity.

Key words: virtual human, Principal Component Analysis(PCA) method, interpolation, extrapolation, time warping, motion capture data, walking engine

中图分类号: 

• TP391