Proper Orthogonal Decomposition of 3D Human Faces
May 13th, 2008 by stankiewicz
The input data (3D point clouds) result from 3D structural light scanning of Virtual Engineering students at PUT. Proper Orthogonal Decomposition has been applied to find the mean (averaged) shape of human face and the most important features (most dominant POD modes) .
Averaged Face
Sample POD modes (positive and negative coefficient value; added to averaged face)
| 57.2% | 13.5% | 5.9% | 4.4% | 4.0% |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
