Weight competitive coding for finger-knuckle-print verification

Abstract

Previous work such as Competitive coding (CompCode) has achieved promising results for online personal authentication based on Finger-knuckle-print (FKP). However, CompCode assigns the same weights for all positions when matching, which will not be stable and will be sensitive to noise, decreasing the performance of matching. In this paper, we propose a new Weighted Competitive coding (W-CompCode) scheme for effective feature matching. In feature extraction stage, we first design a weight matrix for each FKP image based on the Gabor filter response variations at each location. The locations which may have bigger variations will have bigger weights. When matching, the designed weight matrix is incorporated into the angular matching function to measure the similarity between two CompCodes. Furthermore, the weight matrix is also coded and fused with the modified Hamming distance. Experimental results on the PolyU FKP database demonstrate the effectiveness of the proposed method. © Springer International Publishing 2013.