“…Image SR techniques can be applied to regular 2D images and depth images [5,18,20,21,25]. Broadly speaking, within the scope of 2D images, SR tasks can be divided into two categories: multi-image SR and single-image SR. Multi-image SR methods [1,3,8,10,17,24,27] utilize the non-redundant information of multiple frames of the same scene to reconstruct one fine-resolution image. Single image SR, on the other hand, only has one low-resolution input image at disposal which leads to a numerically ill-posed problem.…”