Palmprint recognition based on local Haralick features

In this paper we propose a novel approach to palmprint recognition based on local Haralick features. These features are calculated from the grey-level co-occurrence matrices created on the d × d pixels subimages of the D × D pixels palmprint region of interest (ROI); d <; D defined by overlapping sliding-windows. The biometric template for a person consists of N m-component feature vectors, where N is the total number of subimages and m is the number of Haralick features extracted from a subimage. In order to identify a person, the matching process between the live template and the templates from the system database is performed in N matching modules. Fusion at the matching-score level is used and the final decision is made on the basis of the maximum of the total similarity measure. The results of the recognition and open-set identification experiments are given.