LAUSR

Palmprint-based biometrics has been widely studied for human recognition. However, for feature extraction and matching, most of the current systems use a 2-D image, which can be easily forged. As depth information is included, 3-D palmprints are more competitive in anticounterfeiting. This paper presents a novel person recognition method using 3-D palmprint data. The full-field sinusoidal fringe projection technique is employed to collect 3-D palmprint data remotely and quickly, from which the orientation feature of the mean curvature image is extracted through a revised Gabor filter. An effective feature matching strategy called the binary code list is proposed for classification. Using the developed capture system, a 3-D palmprint database is established, and verification and identification experiments are performed. The PolyU 3-D palmprint database is also used to evaluate the performance of the proposed recognition method. Compared with traditional single-mode feature-based 3-D palmprint recognition methods, the proposed method is more accurate, efficient, and faster.