A Reasearch for Multimodal Finger-Feature Fusion Method Based on Gabor Coding

Extracting robust features have been an important topic in biometrics. For finger-based biometric recognition. Feature extraction is especially important since the finger pose itself is prone to vary during imaging. To reliably represent the multimodal finger-based biometric features, this paper proposed a new feature fusion method using Gabor-based local gray information. Firstly, the Gabor-filtered fingerprint(FP), finger vein(FV) and finger knuckle print(FKP) images were respectively coded by Competitive coding method. Next, the three Gabor-coded images were respectively represented using local-invariant gray description. To represent finger feature globally, these three modal features were fused using a feature series combination strategy. The obtained finger features were called Gabor-based Local-invariant Gray Features(GLGFs). Finally, the experimental results show that the proposed method is capable of achieving higher accuracy recognition in a large homemade database.