Automatic Adjustment of Facial Orientation in 3D Face Image Database

Abstract

The morphometric matching method for identifying persons from surveillance images using a computer-assisted facial image identiˆcation system has now been introduced to actual criminal casework. However, it is still sometimes a laborious task to rapidly distinguish a criminal from a number of suspects using the existing current facial identiˆcation system. Therefore in this study new software, which automatically adjusts the orientation of 3D face images in the database, was designed and its availability was experimentally evaluated as a preliminary step of an automatic face image identiˆcation system. From a total 5000 superimpositions of the 2D face images of 50 subjects which were compared automatically with the 3D face images of 100 suspects, only 18 face images of suspects (0.4) showed inadequate orientation compared to their corresponding 2D face images for recognition purposes. From this experiment, it is concluded that the software developed will be applicable to the automatic adjustment of facial orientation in the 3D face image database and this is anticipated to improved the e‹ciency of policing in Japan as 3D imaging systems are deployed.