Continuous Remote Sensing Image Super-Resolution Based on Context Interaction in Implicit Function Space

Despite its fruitful applications in remote sensing, image super-resolution is troublesome to train and deploy as it handles different resolution magnifications with separate models. Accordingly, we propose a highly-applicable framework called FunSR, which settles unified model by exploiting context interaction within implicit function space. FunSR composes functional representor, interactor, parser. Specifically, the representor transforms low-resolution from Euclidean space multi-scale pixel-wise maps; interactor enables expression global dependencies; parser, parameterized interactor’s output, converts discrete coordinates additional attributes RGB values. Extensive experimental results demonstrate that reports state-of-the-art performance on both fixed-magnification continuous-magnification settings, meanwhile, provides many friendly thanks nature. Our code available at https://github.com/KyanChen/FunSR.