Please enable JavaScript.
Coggle requires JavaScript to display documents.
Robust Solving of Optical Motion Capture Data by Denoising - Coggle…
Robust Solving of Optical
Motion Capture Data by Denoising
Introduction
Errors of Raw optical
motion capture data
High frequency noise or jitter
Occluded markers
Mislabeled markers
Traditional addressing problem
Fix these errors by hand
Extremely time-consuming
Tedious
Purpose of this research
Sidestepping errors of optical motion capture data
producing joing transforms directly from raw marker data
Piece
Starting with a set of marker configurations, and a large database of skeletal motion data such as the CMU motion capture database [CMU 2013b]
Reason of choose
Optical motion capture data :
Denoising motion capture data
CMU dataset
thesis :
Competition : C3D, BVH
joint's global homogeneous transformation matrices
-> local offset (n x m x j x 3)
Not Quaternion
Related work
Motion Capture Cleanup
Markers and joints must follow the laws of physics
Which poses may be possible for the character to achieve and which are unlikely.
Problem : marker swaps
PCA pose-based
Denoising Neural Networks
Pre-processing
Scaling
Scale with uniform height
Scaling factor can either be computed from the T-pose
Representation
Linear Blend Skinning
Local Reference Frame
Statistics