Derivation of 3D-eigenface and facial expression analysis from 3D face form based on eigenspace method

摘要

Facial expression analyses from the luminance images which is the present mainstream have vulnerability to change of the value by illuminating condition and facial direction. Therefore, this paper proposes using "3D-EigenFace" derived by eigenspace method from 3D face form with robustness to them for facial expression analysis, and show its characteristics and experimental result. Moreover, we consider improvement in the rate of discernment by integration with luminance images, and mutual analysis characteristic.