基于关键姿态约束的人体运动序列插值生成
|
摘要
为克服设备精度、捕获速度、位置遮挡、奇异性解析等因素制约,从离散的、有缺失的人体运动捕获数据中生成连贯的人体运动序列,提出了一种基于关键姿态约束的人体运动序列插值生成方法。首先,从人体运动捕获数据库提取各种典型动作对应的关键姿态数据序列,建模为动作节点。其次,依据人体动作过渡关系建立包含所有动作节点的有限状态机图谱。最后,在应用过程中通过最小距离查找即时采集数据对应的动作节点及帧序,若发生数据缺失,则查找缺失数据前后的边界帧所在动作节点,根据运动图谱的过渡关系计算并插入若干过渡动作帧,实现动作的接续连贯。实验结果表明:该方法能有效克服动作数据采集过程的缺失,并可用于复杂人体动作的组合生成,实现动作序列的流畅,提高人体运动捕获数据的可用性、鲁棒性。
Due to device accuracy,capture speed,occlusion and singularity,it remains a difficulty to extract precise and coherent human motions from the discrete and defect data.In this paper,we propose a method that generates human motion interpolation sequences based on key-pose constraint.First,data sequences corresponding to various typical motions are extracted from existing human body motion database,and modeled as motion nodes.Second,a finite state machine map containing all motion nodes is established according to human motion transitions.Finally in the application process,motion node and frame sequences are kept being searched,by using minimum distance,to match the instantaneously captured data.When there are data missing,the motion nodes corresponding to the missing boundary frames are searched,and some transition frames are calculated and interpolated base on the motion map,transferring between motions coherently.The experiments show the method can effectively overcome the deficiency of motion data capture and,it can also be used to combine complex human motions,making smooth motion sequences and improving the usability and robustness of human motion data.
Due to device accuracy,capture speed,occlusion and singularity,it remains a difficulty to extract precise and coherent human motions from the discrete and defect data.In this paper,we propose a method that generates human motion interpolation sequences based on key-pose constraint.First,data sequences corresponding to various typical motions are extracted from existing human body motion database,and modeled as motion nodes.Second,a finite state machine map containing all motion nodes is established according to human motion transitions.Finally in the application process,motion node and frame sequences are kept being searched,by using minimum distance,to match the instantaneously captured data.When there are data missing,the motion nodes corresponding to the missing boundary frames are searched,and some transition frames are calculated and interpolated base on the motion map,transferring between motions coherently.The experiments show the method can effectively overcome the deficiency of motion data capture and,it can also be used to combine complex human motions,making smooth motion sequences and improving the usability and robustness of human motion data.
引文
[1] Xia Shihong,Wei Yi.A Survey of Physics-Based Human Motion Simulation[J].Journal of Computer Research and Development,2010,47(8):1354-1361.
[2] Lee Y,Kim S,Lee J .Data-driven biped control[J].ACM Transactions on Graphics,2010,29(4):1.
[3] CMU.Graphics Lab Motion Capture DataBase[EB/OL].[2017-11-20].http://mocap.cs.cmu.edu/
[4] Wang J,Nie X,Xia Y.Cross-View Action Modeling,Learning,and Recognition[C].IEEE Conference on Computer Vision & Pattern Recognition.IEEE Computer Society,2014.
[5] Jernej Barbic,Alla Safonova.Segmenting motion capture data into distinct behaviors[C].Proc.of Conference on Graphics Interface,2004.
[6] Lucas Kovar,Michael Gleicher.Motion Graphs[J].ACM Transactions on Graphics,2002,21(3):473-482.
[7] Jianyuan Min,Jinxiang Chai.Motion graphs++:a compact generative model for semantic motion analysis and synthesis[J].ACM Trans.on Graphics,2012,31(6):1-12
[8] Daniel Holden.A deep learning framework for character motion synthesis and editing[J].ACM Transactions on Graphics,2016,35(6):1-11
[9] Shihong Xia.Realtime style transfer for unlabeled heterogeneous human motion[J].ACM Transactions on Graphics,2015,34(4):1-10
[10] Biovision.BVH[EB/OL].[2017-10-28].http://research.cs.wisc.edu/graphics/Courses/cs-838-1999/Jeff/BVH.html
[11] Michael Tipping and Christopher Bishop.Probabilistic principal component analysis[J].Journal of the Royal Statistical Society,1999,61(3):611-622
[12] Li Chun-Peng.Motion Synthesis for Virtual Human Using Functional Data Analysis[J].Journal of Software,2009,20(6):1664-1672.
[13] Sun Huaijiang,Xia Guiyu.Survey on Reuse of Human Motion Capture Data[J].Journal of Data Acquisition and Processing,2017,01(32):1-16.
[14] M Firmin,M Van de Panne.Controller Design for Multi-Skilled Bipedal Characters[C].Computer Graphics Forum,2016.
[15] L Liu M van de Panne.Guided Learning of Control Graphs for Physics-Based Characters[J].ACM,2016.35(3):29
[2] Lee Y,Kim S,Lee J .Data-driven biped control[J].ACM Transactions on Graphics,2010,29(4):1.
[3] CMU.Graphics Lab Motion Capture DataBase[EB/OL].[2017-11-20].http://mocap.cs.cmu.edu/
[4] Wang J,Nie X,Xia Y.Cross-View Action Modeling,Learning,and Recognition[C].IEEE Conference on Computer Vision & Pattern Recognition.IEEE Computer Society,2014.
[5] Jernej Barbic,Alla Safonova.Segmenting motion capture data into distinct behaviors[C].Proc.of Conference on Graphics Interface,2004.
[6] Lucas Kovar,Michael Gleicher.Motion Graphs[J].ACM Transactions on Graphics,2002,21(3):473-482.
[7] Jianyuan Min,Jinxiang Chai.Motion graphs++:a compact generative model for semantic motion analysis and synthesis[J].ACM Trans.on Graphics,2012,31(6):1-12
[8] Daniel Holden.A deep learning framework for character motion synthesis and editing[J].ACM Transactions on Graphics,2016,35(6):1-11
[9] Shihong Xia.Realtime style transfer for unlabeled heterogeneous human motion[J].ACM Transactions on Graphics,2015,34(4):1-10
[10] Biovision.BVH[EB/OL].[2017-10-28].http://research.cs.wisc.edu/graphics/Courses/cs-838-1999/Jeff/BVH.html
[11] Michael Tipping and Christopher Bishop.Probabilistic principal component analysis[J].Journal of the Royal Statistical Society,1999,61(3):611-622
[12] Li Chun-Peng.Motion Synthesis for Virtual Human Using Functional Data Analysis[J].Journal of Software,2009,20(6):1664-1672.
[13] Sun Huaijiang,Xia Guiyu.Survey on Reuse of Human Motion Capture Data[J].Journal of Data Acquisition and Processing,2017,01(32):1-16.
[14] M Firmin,M Van de Panne.Controller Design for Multi-Skilled Bipedal Characters[C].Computer Graphics Forum,2016.
[15] L Liu M van de Panne.Guided Learning of Control Graphs for Physics-Based Characters[J].ACM,2016.35(3):29