A robust non-vascular retina recognition system using structural features of retinal image

Abstract

Biometric technology improves the accuracy of the person's identification system instead of the conventional identification technologies such as use of passwords, PIN, token etc. Biometric technologies are automated authentication methods, which identifies person's identity based upon his specific physiological or behavioral traits. Among all biometric systems such as iris, hand vein, finger prints, face, hand geometry, voice, gait, signature etc., human retina provides the most reliable and almost impossible to forge biometric trait. Most of the previous work carried out on retina recognition involves vessel based matching by using feature points i.e. minutiae points. Vessel segmentation and minutiae point extraction is a time consuming process. This motivates us to perform retina recognition matching without using minutiae points. This paper presents a simple and fast non-vascularbased retina recognition system. It computes similarity measure using novel features based upon structural information of an image. It extracts illuminance, contrast and structural features from a color retina image and combines these extracted attributes using an empirically optimized function to generate a similarity score between two candidate images. Finally matching decision is obtained on the basis of highest score value. The proposed system is tested on two retinal image databases collected from local source i.e. RIDB and AFIO. The local databasesare also made available online for other researchers. Efficiency of the proposed system is tested by the computation of false rejection rate (FRR) and false acceptance rate (FAR) and experimental results prove the validity of the proposed system. The method achieves an average identification rate of 92.50% on both databases.