CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleMotion Editing and Synthesis of 3D Human Skeleton Models
Thesis Advisor张晓鹏
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline计算机应用技术
Keyword运动编辑 运动合成 人体骨架模型 运动捕获数据 Motion Editing Motion Synthesis Human Skeleton Model Motion Capture Data
Abstract计算机动画是计算机图形学中一个非常重要的分支领域,其综合了真实感图形生成技术、运动控制原理、图像处理技术等诸多技术。计算机动画的研究对象可以是现实世界和虚拟世界的一切物体,其中最具有挑战性的就是人体动画。三维人体模型的运动编辑与合成技术正是为了研究如何对已有运动数据进行编辑,生成逼真的新运动,从而提高已有运动数据的重用率。这项工作在计算机图形学和计算机动画领域具有重大的研究意义,在三维电影、三维游戏、虚拟现实等领域也具有重要的应用价值。 本文首先对当前三维模型运动编辑的研究现状进行了概述,并对已有的研究方法按照类别划分进行了简要介绍与分析。本文以运动捕获数据为素材,以三维模型特别是三维人体骨架模型的运动为研究对象,进行了如下的研究: 1.提出对单个周期性运动片段的合成方法。为了使新运动更加连贯自然,对骨架模型的拓扑结构进行修改,新结构保留了旧结构中关节点之间的位置关系,并且将支撑脚作为新的根节点,以固定的支撑脚节点为参考,调整模型运动,消除了滑步现象并完成了运动朝向的调整。方法基于几何原理,合成效率高,解决了较难处理的急转弯运动。 2.提出对多个非周期性运动片段的合成方法。将周期性运动的合成扩展到非周期性运动,在运动片段库中搜寻与联接点处运动状态相似的过渡片段,如没有适用结果,则对联结点处运动进行加权生成过渡运动片段;消除运动片段的位移差之后,对过渡区域进行有针对性地动态调整,提高了运动过渡的平滑度。该方法丰富了新动作的类型。 3.建立了一个运动编辑与合成系统。实现对周期性与非周期性运动的联合编辑,系统算法效率高,适用于对运动的实时编辑与合成。 4.在人体骨架模型的运动编辑与合成研究的基础上,对骨架模型运动编辑与网格模型运动的结合进行了进一步的研究。这一探索性的实践对BVH格式的运动捕获数据编辑提供了一个代表性的应用实例。
Other AbstractComputer animation is a very important branch of computer graphics, which concerns many areas in CG, like realistic image illumination, motion control, image processing and so on. The objects of the study can be everything in both real world and virtual world, among which the human animation is one of the most challenging ones. The motion editing and synthesis of 3D human model focus on how to edit the existing motion data to generate vivid new motion. The reusing rate of these data can also be improved by this work. This study has not only great academic sense in both computer graphics and computer animation, but also important application value in 3D movie, 3D video game and virtual reality. In this thesis, we present the state of the art in the research of motion editing of 3D models. Then we briefly introduce and analyze these existing algorithms according to the categories. The main work is carried out as follows, in which the motion capture data is the original material and the 3D models, especially the human skeleton models, are the objects of study: 1.We propose an efficient method for periodical motion synthesis. In order to make the new motion continuous and natural, we modify the given topological structure of motion capture data. New structure saves the position relations between joints and adjusts the motion according to the new root joint, the supporting foot. By doing this, the foot-slide problem is solved and the motion direction is modified. Based on geographic theory, this method has high processing efficiency and can resolve the motion at sharp turning. 2.We propose an efficient method for nonperiodic motion synthesis. In this method, we search for the proper transition motion segment with similar posture compared with the posture at junction in the motion data library. If no similar segment is found, we generate the transition segment by using weighted sum formula on junction motions. After eliminating the position offset, we dynamically adjust the transition range to improve the smoothness. This method enriches the types of new motions. 3.We build a motion editing and synthesis system, in which both periodical and nonperiodic motions are processed. This system can be used in real-time application because of the high efficient algorithm. 4.We try to transplant the motion editing method of skeleton models to the motion of mesh models. This work puts the method into practice and provides a typical application of motion capture da...
Other Identifier200828014628034
Document Type学位论文
Recommended Citation
GB/T 7714
郭晓乐. 三维人体骨架模型的运动编辑与合成[D]. 中国科学院自动化研究所. 中国科学院研究生院,2011.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20082801462803(7761KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[郭晓乐]'s Articles
Baidu academic
Similar articles in Baidu academic
[郭晓乐]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[郭晓乐]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.