Satellite and Aerial Image Mosaicing - A Comparative Insight

Abstract

Image registration or image stitching is a central operation in many useful and important tasks in image processing like maps construction, scanning large documents and panoramic photos creation. In particular image mosaicing is used to assemble several overlapping images in order to constitute the global frame. We will focus on a feature-point matching method to perform the mosaicing. The SIFT algorithm is used to extract the feature points in both images. The mosaicing result is obtained after transforming the sensed or target image to align to the reference image. Performing a mosaicing operation is not sufficient to claim reaching the goal. Objective metrics must be used to evaluate the resulting mosaic. In this paper we present a complete mosaicing system named EsiReg and give a brief comparative insight on results of stitching satellite and aerial images using well known performance metrics.