Abstract
Image registration is an important step in all image processing fields that require the integration of information from multiple observations into a single image. Image registration is a fundamental inverse problem in imaging. A large number of researchers have worked in the field of image registration and several registration methods have been proposed. In image registration, the relative shifts between low-resolution (LR) images compared to a reference LR image are estimated with fractional pixel accuracy. Images of the same scene are acquired at different times, often on a regular basis, and possibly under different conditions. The sensed image is transformed by means of the mapping functions. Area-based methods are able to make the registration of images in the case of a small shift or rotation between the images. Feature-based matching methods are typically applied when the local structural information is more significant than the information carried by the image intensities.