摘要
针对传统的虚拟人动画生成存在的实时性差、逼真度低、动画生成复杂、成本高等问题,提出了一种基于Kinect体感捕捉设备的实时虚拟人动画生成方法。利用正向运动学方法建立具有拓扑骨架结构的虚拟人模型,将Kinect读取到的用户骨骼信息代入正向运动学方程进行求解,并重定向到虚拟人模型,驱动模型做出真实的一致性运动。为了消除虚拟人运动中的悬浮、穿透和部分关节点丢失现象,提出了平移补偿和关节点丢失保持机制。实验结果表明,该方法能实时逼真的驱动虚拟人运动并生成动画,且该方法易于实现、实用性强。
Traditional virtual human animation has some drawbacks including low real-time performance,low realism,complex generation process and high cost. This paper proposes a novel generation approach to virtual human animation based on Kinect. Firstly,it uses the forward kinecmatics method to establish the skeleton model with topology structure,and then remap the user＇s skeleton information captured by Kinect to virtual human. Furthermore,it introduces the principle of translation compensation and distortion processing mechanism. The experiment shows the method can provide real-time and realistic virtual human animation,and is easy to realize.
出处
《电子科技》
2015年第7期149-152,共4页
Electronic Science and Technology
基金
浙江省自然科学基金重点资助项目(Z1091077)
