Abstract
In this paper, digital image correlation (DIC) and a multi-camera system are being used to reconstruct 3D shapes of different parts of the human face and then aligning these parts using the Iterative Closest Point (ICP) algorithms to produce the full 3D shape of the human face. A simple method to remove the more sparse 3D point cloud in the overlapping area between two 3D shapes is developed by iteratively using the principal component analysis method (PCA). In addition, displacement measurements on the surface of the skin that are caused by a facial expression are measured. The experiment apparatus and set up are seven DSLR cameras that are mounted on a half circle light metal pipe and three tripods. A practical method for calibrating this system is developed by using a designed double sided calibration board, with check marks on both sides that are aligned with each other. A safe and practical method to apply a random speckle pattern on a human's face is developed by using an air brush make up set.
