人体运动的合成与控制是计算机图形学研究的热点之一,但由于人体的关节变量很多且人类对自身运动非常熟悉,使得这项研究变得极具挑战性.为了解决该领域存在的问题,很多研究者尝试通过对运动数据进行降维在低维空间中对人体运动进行合成,或是利用低维信号进行运动控制,均取得了很好的效果,当然也有各自的局限性.文中以运动数据的低维表达为主线,回顾和总结了人体动画领域一些最新的研究成果,并对基于低维数据的人体动画生成技术的发展趋势进行了展望.
Human motion synthesis and control is an important research topic in computer graphics. Because of a large number of joints in a human body and the familiarity of people to their own movements, this problem is very challenging. In order to solve this problem, many researchers try to synthesize human motion in low-dimensional space by reducing the number of dimensionality of motion data or controlling the character movement by low-dimensional signal. Both of these techniques have achieved good results, but also with respective drawbacks. We review and summarize the latest research results of character animation in the reduced low-dimensional space, and also discuss the future trends of human animation generation on the basis of low-dimensional data.