In personal authentication systems (PAS) using fingerprints, there is a decline in authentication accuracy because of the dirt on the contact surface and a problem on which a contact surface deteriorates with the present input equipment. We have already examined the possibility of non-contact fingerprint authentication to address these problems. In this paper, we propose an improvement for the extraction of a fingerprint and a compensation method for a rotation error. In addition to information using luminosity, color is also used to improve the fingerprint extraction technique. It notes the valley line of a fingerprint is bluer than a ridge line. Emphasis of the information of a principal line (valley line) to the region which is disconnected when the conventional technique is used, is effective to control the disconnection. A principal line is enhanced using the information on Z component of the XYZ color system which is a blue component. Moreover, if a non-contacting input is carried out, three kinds of rotation errors, yaw, roll, and pitch will occur simultaneously. Then, we have to expand the compensation of a rotation error from two dimensions to three dimensions. In the proposed rotation compensation technique, the front view and side view of a fingertip are used. Yaw, roll and pitch angles are computed from the shape feature of a finger, and it compensates using affine transformation. As a result of the experiment, the thinning noise decreased and the quality of the principal line improved greatly
展开▼
