A new robust fuzzy fusion technique in super resolution imaging

摘要

Key objective of super-resolution (SR) is to overcome the ill-posed conditions of image acquisition. SR facilitates better content visualization and scene recognition from low resolution images. In this paper we present a new robust super resolution approach. Our approach, firstly registers two input image using SIFT-BP-RANSAC registration. Secondly due to the importance of information gain ratio of SR, the fuzzy inference system will fuse the registered image with the reference image to aggregate the amount of details from both input images. According to the diversity of acquisition model classes, an adaptive fuzzy approach has been developed to robustly fuse the integrated information of input images into a high resolution image. Our approach iteratively minimizes the difference between the resulted high resolution image and the ground truth image. Independence from the acquisition model leads to the robustness of our method on different ill-posed capturing conditions. Our final results indicate better achievements in comparison with similar recent works in the literature.