Image Registration Techniques and Frameworks

Abstract

Image registration allude to transforming one image with reference to another (geometrically alignment of reference and sensed images) i.e. the process of overlaying images of the same scene, seized by assorted sensors, from different viewpoints at variant time. Virtually all large image evaluating or mining systems require image registration, as an intermediate step. Over the years, a broad range of techniques has been flourished for various types of data and problems. These approaches are classified according to their nature mainly as area-based and feature-based and on four basic tread of image registration procedure namely feature detection, feature matching, mapping function design, and image transformation and resampling. The current chapter highlights the cogitation effect of four different registration techniques, namely Affine transformation based registration, Rigid transformation based registration, B-splines registration, and Demons registration. It provides a comparative study among all of these registration techniques as well as different frameworks involved in registration process.