Automatic biometric verification algorithm based on the bifurcation points and crossovers of the retinal vasculature branches. (27th January 2020)

• Record Type: Journal Article
• Authors: Betaouaf, Talib Hichem; Decencière, Etienne; Bessaid, Abdelhafid
• Is Part Of: International journal of biomedical engineering and technology. Volume 32:Number 1(2020)
• Page Start: 66
• Page End: 82
• Publication Date: 2020-01-27
• Subjects: biometrics -- biometric verification -- retinal blood vessel -- image segmentation -- bifurcation points; Biomedical engineering -- Periodicals; 610.28
• Languages: English
• ISSNs: 1752-6418

Abstract

Biometric identification systems allow for the automatic recognition of individuals on one or more biometric characteristics. In this paper, we propose an automatic identity verification algorithm based on the structure of the vascular network of the human retina. More precisely, the biometric template consists of the geometric coordinates of bifurcation points and crossovers of the vascular network branches. The main goal of our work is to achieve an efficient system while minimising the processing time and the size of the data handled. Therefore, this algorithm uses a novel combination of powerful techniques for feature extraction based on mathematical morphology, like the watershed transformation for the segmentation of the retinal vasculature and hit-or-miss transform for the detection of bifurcation points and crossovers. We detail each step of the method from acquisition and enhancement of retinal images to signature comparison through automatic registration. We test our algorithm on a retinal images database (DRIVE). Finally, we present and discuss the evaluation results of our algorithm, and compare it with some of the literature.