Conjunctival Vasculature (CV) as a unique modality for authentication, using Steady Illumination Colour Local Ternary Pattern (SIcLTP)

Abstract

it has been proved that a new biometric modality based on the patterns of conjunctival vasculature performs well in visible spectrum. The vessels of the conjunctiva could be seen on the visible part of the sclera; these vessels are very rich and contain unique details in the visible spectrum of light. In this paper we have explored the feature extraction technique for conjunctival vasculature using Steady Illumination colour Local Ternary Patterns(SIcLTP). The concept of LTP as argued in various earlier published papers is that, it is very robust to noise and gives rich information at the pixel level. In this paper before feature extraction the images are converted into YIQ colour space from RGB colour space to do away with the redundant information demonstrated by RGB colour space. Further the image similarity and dissimilarity is found out using zero-mean sum of squared differences between the two equally sized images. The results received with AUC (Area Under ROC Curve) being 0.947, demonstrates the richness of the texture pattern of conjunctival vasculature and robustness of the method being used. It is concluded that this texture pattern is a very promising biometric modality which could be used for identification.