A Complete Feature Matching Fusion Method for Palmprint Recognition

As we all know, the single model extraction palmprint biometric recognition information is relatively simple, and the existing traditional palmprint recognition methods are greatly affected by the external environment. In this paper, we propose a fusion method of traditional methods and deep learning, fusing at the feature layer and matching value layer respectively. The fusion strategies are maximum fusion and summation fusion. The experimental results achieve high recognition accuracy and low Equal Error Rate on both constrained environment and unrestricted environment data sets.