Critical Analysis of Image Descriptors

Abstract

Image description is a sub field of image processing. Local features or nearby elements and their descriptors are robust and compact vector representations of their neighborhood, and building squares of numerous computer vision applications. Descriptors are picking up prominence for getting correspondence between two pictures, and seeking and finding the required picture from dataset. This paper presents the detailed comparison & experimental analysis of different existing image description techniques such as Speeded up Robust Features,Binary Robust Invariant Scalable Keypoints, and Fast Retina Keypoint. Their performance is analyzed on MATLAB (2015a). In this research study three image descriptor algorithms are analyzed in terms of their detected feature points, considering two scenarios (Scenario 1: simple image and Scenario 2: cluttered scene) and two cases (Case 1: Rotation and Case 2: Scaling) along with the image correspondence computation in three cases (Simple, Rotation and Scaling