Abstract:

In order to solve the problems of few detail information and relatively high false-rejection and falseacceptance rates in vein recognition, a new fusion algorithm of the weighted information from knuckleprint and vein is presented. In the preprocessing stage of knuckleprint, an 8-neighborhood method is used to accurately locate the basic points of palm, and the straight-line fitting technique is adopted to fit the contour of finger and to further locate the region of interest （ROI） of knuckleprint. In the matching stage of image, the characteristics of knuckleprint are extracted using a Gabor filter and is then fused with vein information via the weighted fusion technology. Simulated results indicate that the proposed algorithm improves the recognition rate and system stability.

Key words: vein recognition, knuckleprint, straight-line fitting, Gabor filter, weighted information fusion