A Novel Hybrid Multibiometrics Based on the Fusion of Dual Iris, Visible and Thermal Face Images

Human identification via multibiometrics is a very promising approach to improve overall systems security and recognition performance. In this paper, a novel multibiometrics fusion strategy based on dual iris, visible and thermal face traits is proposed. Firstly, 1D Log-Gabor dual iris codes are fused into a serial feature vector to address the noisy problems of iris. Then, Complex Gabor Jet Descriptor (CGJD) is proposed to enhance the feature representation abilities for thermal and visible face images. Finally, a score level fusion algorithm based on Aczél-Alsina triangular norm (AA t-norm) is proposed to enlarge the distance of imposters and narrow the distance of genuine simultaneously. All the simulations are performed on a virtual multibiometrics database, which merges the challenging CASIA-Iris-Thousand database with dual iris and the NVIE face database with visible and thermal face images. The proposed fusion strategy achieves EER of 2.89×10^-4 and d of 8.87, which outperforms the state-of-theart methods.