Palmprint Recognition Method Based on Adaptive Fusion

Abstract

Bimodal biometrics can overcomes some kinds of limitations of single biometrics and obtain a higher accuracy than single biometrics. In this paper, we propose a palm print recognition method based on the adaptive fusion of 2D and 3D palm print images. 3D palm print contains the depth information of the palm surface, while 2D palm print contains plenty of textures. Firstly, the biometric trait can be obtained by an adaptive fusion method. Combine the 2D and 3D Palm print images together by a complex vector. In this phase, we use the automatic weighted combination strategy. We assume that any test sample can be expressed as a linear combination of all the training samples in complex space. Then we can find M near neighbors of the test sample by solving the linear system and use the effect of the M near neighbors to perform classification. The experimental results show that the proposed method can obtain a higher accuracy.