Creating Personal 3D Avatar from a Single Depth Sensor
Authors:
- Wenjuan GongWGWenjuan Gong
- Zhichao WangZWZhichao Wang
- Weishan ZhangWZWeishan Zhang
- 1 August 2015
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1193-1198
Abstract
Personalized avatar provides splendid user experiences in gaming, virtual community and other applications. The emergence of commodity depth sensors, e.g., The Kinect device, provides a wealth of input for personalizing avatar. While there are a wide range of research done on creating personal 3D avatar that looks exactly like the user, how to personalize an animated character, like Shrek, to look like the user is seldom explored. In this paper, we propose a method for creating 3D avatars with similar body shape as the user. First, we capture a 3D human body model of the user, and the user specifies an animated character. Then the captured user model is simplified and key points are selected from both the user model and the animated character. Finally, all key points are transformed directly form the animated character to the personalized model and other points are interpolated with RBF interpolation. We show an example of a user model transformed into a warrior avatar.Keywords
This publication has 10 references indexed in Scilit:
- Tensor Body: Real-Time Reconstruction of the Human Body and Avatar Synthesis From RGB-DIEEE Transactions on Cybernetics, 2013
- KinectFusion2011
- Real-time human pose recognition in parts from single depth images2011
- Articulated pose estimation with flexible mixtures-of-parts2011
- A Statistical Model of Human Pose and Body ShapeComputer Graphics Forum, 2009
- OBJCUT: Efficient Segmentation Using Top-Down and Bottom-Up CuesIEEE Transactions on Pattern Analysis and Machine Intelligence, 2009
- A discriminatively trained, multiscale, deformable part model2008
- 3-D reconstruction of static human body shape from image sequenceComputer Vision and Image Understanding, 2004
- The space of human body shapes2003
- Surface simplification using quadric error metrics1997