Protection of Sparse Retinal Templates Using Cohort-Based Dissimilarity Vectors

Retinal vasculature is a biometric characteristic that highly accurate for recognition but which no template protection scheme exists. We propose the first retinal scheme, adapting an existing paradigm of cohort-based modelling to templates containing node and edge data graphs. The results in at most 2.3% reduction accuracy compared unprotected templates. A common concern with cohort based systems availability distance scores can be exploited reconstruct image or using inversion attack. On contrary, we show our sparse system less than 0.3% success rate inverse In addition, rigorous unlinkability analysis shows has linkability least as low lower state-of-the-art schemes.