A Motion Segmentation Based Algorithm of Human Motion Alignment
As a classic method of human motion alignment, original Dynamic Time Warping algorithm may cause poor effects while applied in the field of human motion blending, such as motion pause and motion distortion. To address this problem, we propose a motion seg
- PDF / 250,749 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 7 Downloads / 213 Views
Digital Media and Human Computer Interaction Research Center, Hangzhou Normal University, Hangzhou, China [email protected], [email protected] 2 School of Computer Science and Technology, Zhejiang University, Hangzhou, China [email protected]
Abstract. As a classic method of human motion alignment, original Dynamic Time Warping algorithm may cause poor effects while applied in the field of human motion blending, such as motion pause and motion distortion. To address this problem, we propose a motion segmentation based method of human motion alignment. We first make use of Isomap which is a classic algorithm in manifold learning to reduce dimensionality of all the motion frames in the original motion samples. Then we can obtain the segment of different motion samples by making use of the extreme point of all the corresponding low-dimensional coordinates of motion frames. Finally we apply Dynamic Time Warping method in the corresponding segments of different motion samples respectively to achieve refined alignment results. We conduct contrast experiment between our algorithm and traditional method, the result shows that our method can eliminate motion pause and motion distortion phenomenon caused by traditional method and make significant improvement in the application of human motion blending. The generated motion samples have a high degree of both fidelity and fluency. Keywords: Motion alignment Isomap Motion segment
Motion blending
Dynamic
time warping
1 Introduction The control and synthesis of 3D human motion have been one of the hot spots in the research field of computer graphics all along. And with motion capture technology being widely used in the field of digital entertainment and system simulation, data driven human motion technology has increasingly become an important research topic in the field of computer animation. Benefiting from many large-scale 3D human motion database been established, the importance of mining the value of existing motion capture data and make good use of them is becoming increasingly prominent. Motion data pre-processing has always been the previous step of reuse of motion data. Motion data pre-processing also includes a variety of forms, such as: motion denoising, motion space warping, motion alignment, etc. The major purpose of motion denoising is © Springer Science+Business Media Singapore 2016 L. Zhang et al. (Eds.): AsiaSim 2016/SCS AutumnSim 2016, Part II, CCIS 644, pp. 660–670, 2016. DOI: 10.1007/978-981-10-2666-9_67
A Motion Segmentation Based Algorithm of Human Motion Alignment
661
to eliminate the generated noise data which are caused by the defects of motion capture equipment, the commonly used methods include linear time invariant filter [1, 2], the Kalman filter [3–5] and some data driven filter [6, 7]. Motion space warping is mainly in order to adjust the spatial coordinates of the moving segment or the whole direction, so that the data can be processed in a unified coordinate space [8]. Motion alignment aims to find the temporal correspondence between diffe
Data Loading...
