An analysis of a robust super resolution algorithm for infrared imaging

This paper considers the performance of a robust super resolution (SR) algorithm on simulated infrared images containing scenes with very different textures-corresponding to different geographical textures (urban and rural). The aim is to assess the usefulness of the robust SR technique for the detection of small objects in infrared imagery-in the presence of typical image noise and variations in the background scene. The need for accurate (sub-pixel) image registration is stressed and candidate algorithms are proposed.