Evaluation of Feature Descriptor on D-Saddle Keypoint Detection in Retinal Image Registration

Abstract

This paper presents an investigation of the common feature descriptor on D-Saddle keypoint detection to guide the retinal image registration process. Few feature descriptor methods have been chosen which are Histogram of Oriented Gradient (HOG), Speeded-Up Robust Features (SURF), Binary Robust Invariant Scalable Keypoints (BRISK) and Fast Retina Keypoints (FREAK). Findings indicate that the combination of SURF descriptor with D-Saddle detection produced more significant inliers as compared to the other state-of-the-art combinations of descriptors and keypoint detections.