Multimodal system using Radon-Gabor transform
Milind Eknath Rane, Umesh S. Bhadade · 2017
Reliable person recognition schemes are demanding in variety of applications, either to confirm or determine the identity of an individual. This is still a challenging problem, addressed by many researchers using unimodal biometric. However, research shows that the performance of unimodal biometric systems is limited. To improve the performance, in the proposed work, multiple biometric modalities of a person are used. Particularly, face and palmprint are used as a biometric trait for the person recognition task. Multiresolution features based on Radon Gabor transform are proposed in this work. As the Radon transform is invariant to rotation and a Gabor transform provides the multiple orientations, the proposed feature improves the recognition rate up to 98%. At feature level, the fusion of face and palmprint is used. The Performance of the proposed system is evaluated on IITD, PolyU_Palmprint and Face 94 databases.