Multi-Sensor Image Registration for Remote Sensing under Scale Invariant Feature Transformation

Abstract

Image registration deals with establishing correspondence between pictures of an equivalent scene or object. A picture registration rule ought to handle the variations introduced by the imaging system capturing the scene. Scale Invariant Feature remodel (SIFT) is a picture registration rule supported native options in a picture. Compared to the previous registration algorithms, SIFT is a lot of sturdy to variations caused by changes in size, illumination, rotation, and viewpoint of the pictures. As a result of its performance, the rule is wide studied, modified, and with success applied in several image and video primarily based applications, within the domains akin to drugs, industry and defense. This paper is associate outcome of in depth study on the state-of-art image registration algorithms supported SIFT. However, directly applying SIFT to remote sensing image registration usually ends up in a really variety of feature points or key points, however, a tiny low number of matching points with a high warning rate. We tend to argue that this is often because of the actual fact that spatial data is not thought about throughout the SIFT-based matching method. This paper proposes a way to enhance SIFT-based matching by taking advantage of neighborhood data. The planned methodology generates a lot of correct matching points because the relative structure in numerous remote sensing pictures area unit virtually static.