Image Registration for Super-Resolution

Introduction

Super-resolution reconstructs high-quality, high-resolution images by exploiting the fact that due to the relative motion between the camera sensor and the true scene, each aliased, under-sampled low-resolution frames acquired contains distinct incomplete and degraded scene information about the true scene. Therefore, in order for super-resolution to acquire this distinct scene information and successfully generate a high-resolution image, accurate knowledge of registration parameters is required for each of the input low-resolution frames. To formally define image registration, it is the process of geometrically aligning two or more images taken at different times, from different viewpoints and/or by different sensors. In various computer vision applications such as remote sensing, medical imaging, target detection, super-resolution imaging and many more, image registration is the most crucial intermediate process. To achieve accurate super-resolution image reconstruction, it is critical for image alignment to be precise. Mis-alignment of the under-sampled low-resolution frames will result in the reconstruction of an erroneous high-resolution image which may not be a true approximation of the original scene.