Differential Thresholding of SIFT-based Blood Vessel Segmentation in Retinal Fundus Image

Automatic vessel segmentation often makes the worse performance on extraction factor since it is desirable to have continuous structure. Henceforth, imperative to fill and examine the gaps among vessels. This paper has spotted the limitations of two different segmentation models and practiced a new segmentation model that incorporates SIFT method followed by differential thresholding. SIFT method discovers the unique key points that cannot be varied by any scaling or rotation process. This proposed segmentation process skips overseeing the vessel portions as major and minor. Instead, the proposed work evaluates the vessel portions based on their scale invariant characteristics and pixel intensity. Finally, the proposed model compares its performance over other conventional segmentation models including Iterative Segmentation model and Gaussian Mixture Model (GMM) in terms of Accuracy, Specificity, Sensitivity and Area Under the Curve (AUC) and proven the efficiency of proposed work.