Face identification using a 3D gray-scale image-a method for lessening restrictions on facial directions

We propose a method to identify the human face using a 3D gray scale image which combines a 3D image with a gray scale image. The proposed method can identify facial images which face in various directions. First, 3D positions, where the eyes and the tip of the nose are located, are estimated in the acquired 3D gray scale image. Next, the calibration of facial directions and brightness is performed. Due to facial images facing more to the right or the left, some parts of the calibrated data are lost by occlusion. Thus, we assume that a human face is almost symmetrical, and the side of the facial image without any loss of data is regarded as the region for identification. Feature vectors which reflect individuality are extracted from this region. The identification procedure is conducted using the subspace method. In order to demonstrate the efficiency of this method, the experiment used 3D gray scale images of 24 people.