1Associate Professor, Department of Biomedical Engineering, Jerusalem College of Engineering, Chennai, India
2Third Year Students, Department of Biomedical Engineering, Jerusalem College of Engineering, Chennai, India
*Corresponding Author: J. Sivakamasundari Associate Professor, Department of Biomedical Engineering, Jerusalem College of Engineering, Chennai, India Email: sivakamasundarij17@gmail.com
Online published on 19 August, 2019.
The retinal biometric system uses automatic segmentation of vasculature in retinal images. The Harmony Search Algorithm (HSA) combined with conventional Multilevel Thresholding (MLT) methods are used to produce continuous vasculature information for the application of retinal biometric system. Publically available retinal image database is employed for this work. The preprocessed retinal images are subjected to HSA based Otsu MLT methods by the best objective functions. The validation is carried out with segmented images against its ground truth images using binary similarity measures. From the segmented images features like statistical, textural and structural are analyzed. The matching is carried out for query images with already enrolled images using Euclidean Distance measures and decision is made. This system gives 96% matching efficiency. Hence this algorithm could be recommended for retinal biometric system.
Braille, retinal biometric system, retinal fundus image, blood vessels, harmony search algorithm, multilevel thresholding, segmentatione
