Abstract
In this paper, the authors describe an automatic system for identifying individuals by the facial information obtained by using the three dimensional sensor. The authors have proposed a method for identification of human facial data obtained by using the fiber grating vision sensor which has been developed by the authors. But this method has two problems of the following; the strict restrictions on a subject person at the time for obtaining a facial data; and the coarse sampling range data obtained by using the fiber grating vision sensor. First, the authors propose a method for obtaining three dimensional facial data by using the fiber grating vision sensor. The present method is able to obtain the facial data without forcing the subject person to put his face at a fixed position and a fixed direction exactly. In this system, a mark on a floor and a mirror in front of the fiber grating vision sensor are used for the restrictions on the subject person. The useful data for the identification is obtained with forcing the subject person only to stand on the mark and to reflect his tip of nose in the mirror. Next, the authors propose a method for identifying the human faces accurately without being affected by the coarse sampling range data. In this system, to identify human faces, a multi-layered neural network is used in which the inputs are two component values of normal vector at data spot in the facial surface. To eliminate the effect of the coarse sampling range data is carried out by planing in learning process of the neural network. In this paper, the experiments using the experimental system are performed to demonstrate the efficacy of this system and the experimental results are shown.
