Three Dimensional Palmprint

hand shape, etc. Two dimensional (2D) palmprint recognition [1-4, 16-18] has been widely studied in the past decade and it has been proven that palmprint is a unique biometric identifier. 2D palmprint systems have merits of high accuracy and user friendliness, etc. Nonetheless, 2D palmprint can be easily counterfeited and much three dimensional (3D) palm structural information is lost. Inspired by the success of 3D techniques in biometric authentication, such as 3D face [5] and 3D ear recognition , very recently a structured-light imaging [7, 8] based 3D palmprint system [9] was developed to capture the depth information of palmprint. In [9], the Mean curvature and Gaussian curvature are calculated from the depth information and they serve as the basic features for 3D palmprint matching and recognition. As shown in [9], the Mean curvature is a stable and distinct feature of 3D palmprint. By normalizing and mapping the Mean curvature values to a plane, we can get a Mean Curvature Image (MCI) which contains line structure features and texture features of the 3D palmprint. In [9], the MCI was binarized to highlight the line features and the binarized MCI was used as the feature map for 3D palmprint matching. However, the binarization operation loses much the texture information existing in the MCI. Actually, if we view the MCI of a 3D palmprint as a 2D palmprint image, then many 2D palmprint feature extraction techniques can be applied. In addition, the line and texture features could provide complementary information for palmprint discrimination. Therefore, in this paper we propose to extract both line