Local and Global Skeleton Fitting Techniques for Optical Motion Capture
CAPTECH '98 Proceedings of the International Workshop on Modelling and Motion Capture Techniques for Virtual Environments
Mapping optical motion capture data to skeletal motion using a physical model
Proceedings of the 2003 ACM SIGGRAPH/Eurographics symposium on Computer animation
Self-Calibrating Optical Motion Tracking for Articulated Bodies
VR '05 Proceedings of the 2005 IEEE Conference 2005 on Virtual Reality
| Hi-index | 0.00 |
Motion Capture is a widely accepted approach to capture natural human motion, usually utilizing markers to track certain anthropological points on the participant's body. Unfortunately, these markers do not carry any identification information. Furthermore, marker data can be noisy. To address these problems this work suggests a hybrid approach, i.e. an approach using several experts to solve easier, less complex subproblems. Currently, the presented hybrid approach is built upon three methods, two for identification and one for tracking purposes. For identification of an initial posture, a PCA-based technique for aligning a skeleton model as well as a tree-based optimization comparing anthropometric and tracking data are introduced. To complement the hybrid computation pipeline a neural network algorithm based on self-organizing maps tracks the markers on subsequent frames.