Tracking People with Twists and Exponential Maps
CVPR '98 Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
Capturing and animating skin deformation in human motion
ACM SIGGRAPH 2006 Papers
As-rigid-as-possible surface modeling
SGP '07 Proceedings of the fifth Eurographics symposium on Geometry processing
Data-driven modeling of skin and muscle deformation
ACM SIGGRAPH 2008 papers
Realtime performance-based facial animation
ACM SIGGRAPH 2011 papers
KinectFusion: real-time 3D reconstruction and interaction using a moving depth camera
Proceedings of the 24th annual ACM symposium on User interface software and technology
Template deformation for point cloud fitting
SPBG'06 Proceedings of the 3rd Eurographics / IEEE VGTC conference on Point-Based Graphics
Hi-index | 0.00 |
Modeling realistic skin deformations due to underneath muscle bulging has a wide range of applications in medicine, entertainment and art. Current acquisition systems based on dense markers and multiple synchronized cameras are able to record and reproduce fine-scale skin deformations with sufficient quality. However, the complexity and the high cost of these systems severely limit their applicability. In this paper, we propose a method for reconstructing fine-scale arm muscle deformations using the Kinect depth camera. The captured data from the depth camera has no temporal contiguity and suffers from noise and sensory artifacts, and thus unsuitable by itself for potential applications in visual media production or biomechanics. We process noisy depth input to obtain spatio-temporally consistent 3D mesh reconstructions showing fine-scale muscle bulges over time. Our main contribution is the incorporation of statistical deformation priors into the spatiotemporal mesh registration progress. We obtain these priors from a previous dataset of a limited number of physiologically different actors captured using a high fidelity acquisition setup, and these priors help provide a better initialization for the ultimate non-rigid surface refinement that models deformations beyond the range of the previous dataset. Thus, our method is an easily scalable framework for bootstrapping the statistical muscle deformation model, by extending the set of subjects through a Kinect based acquisition process. We validate our spatio-temporal surface registration method on several arm movements performed by people of different body shapes.