Palmprint recognition based on multiple feature information fusion

In this paper, we propose an information fusion-based palmprint recognition method. Firstly, three different kinds of palmprint features including texture, principal line and appearance features are extracted using 2D Gabor phase encoding, the modified finite Radon transform (MFRAT) and orthogonal discriminant locality preserving projections (ODLPP), respectively. Then, Hamming distance, pixel-to-area comparison-based similarity calculation and Euclidean distance are presented to computing their matching score, and the nearest neighborhood (NN) classifier is applied to classification. Lastly, information fusion on the decision level is conducted to further improve the recognition accuracy. The experimental results on the PolyU palmprint database demonstrate the feasibility and effectiveness of the proposed approach.