Abordagem baseada em morfologia matemática para a identificação biométrica através da iris

Abstract

This paper presents a proposal for biometric identification of individuals by analyzing the iris, using an approach based on Mathematical Morphology. Algorithms based on morphological operators are used in several stages of processing, segmentation, enhancement of patterns, extraction and representation of features. To evaluate the algorithms, tests were performed using images of the University of Bath Iris Image Database - UBIID, of the Chinese Academy of Sciences Institute of Automation - CASIA, and acquired images to compose the Brazilian Iris Research Database - BIRD. In order to assess the performance of classification, the proposed method was compared to Daugman’s method (reference). The results indicate the viability of the proposed approach for iris recognition. For UBIID images, the performance of the proposed method was superior to the reference method, achieving 99.35% accuracy at the point of equal error (EER - Equal Error Rate), where the false acceptance rate (FAR) and false rejection rate (FRR) are equal.