Single Image Super-Resolution


Single image super-resolution (SISR) is a fundamental low-level computer vision problem that aims to recover a high-resolution image from its low-resolution counterpart. There are two main reasons for performing SISR: (i) to enhance the visual quality of an image for human consumption, and (ii) to improve the representation of an image for machine perception. SISR has many practical applications including robotics, remote sensing, satellite imaging, thermal imaging, medical imaging, and much more. In perception systems, images are represented as 2D arrays of pixels whose quality, sharpness, and memory footprint are controlled by the resolution of the image. Consequently, the scale of the generated high-resolution image is fixed depending on the training data. Thus, instead of training multiple models for various resolutions, it can be extremely useful in terms of practicality to have a single SISR architecture that handles arbitrary scale factors. This is especially true for embedded vision platforms (e.g., unmanned ground/aerial vehicles) with multiple on-board cameras that must execute difficult tasks using limited computational resources.


  • We introduced a novel implicit neural network for the task of single image super-resolution at arbitrary scale factors


  1. Q.H. Nguyen and W.J. Beksi, "Single Image Super-Resolution via a Dual Interactive Implicit Neural Network," IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2023.
    Paper  •   Supplement  •   Preprint  •   Source Code  •   Citation