Perturbation-enhanced feature correlation filter for robust iris recognition

Visual pattern of human iris provides rich texture information for personal identification. However, it is challenging to match intra-class iris images with large variations in applications. This study proposes a perturbation-enhanced feature correlation filter (PFCF) for robust iris matching. PFCF is developed based on quad-phase minimum average correlation energy filter, but it has two significant improvements. First, PFCF is performed on Gabor filtered iris images to encode both local and global features. On the one hand, Gabor images can enhance the local details of iris texture. On the other hand, correlation filters describe the regional appearance information and measure the global similarity between iris images efficiently. Secondly, artificially perturbed iris images are generated to model intra-class variations. Also, a set of additional correlation filters are developed accordingly as the gallery templates. The decision is determined by the fusion result of multiple correlation filters. Therefore PFCF not only takes the advantages of Gabor images and correlation filters but also enlarges the amount of enrolled templates for robust iris matching. Extensive experiments on three challenging iris image databases demonstrate that the proposed method outperforms the state-of-the-art methods according to its robustness against deformation, rotation, occlusion, blurring and illumination changes in iris images.