Output details
11 - Computer Science and Informatics
Lancaster University
Single-Image Super-Resolution Using Sparse Regression and Natural Image Prior
<26> Image enhancement is fundamental to many industries, but requires domain-specific expertise. Our work, which is now cited as a standard reference for learning-based image enhancement, removes this requirement by allowing users to build custom image enhancement software from examples without having to acquire domain specific expertise. We solve this problem with a novel, fast way to learn models from millions of image pairs. This has high impact on single-image super-resolution and image denoising applications, and applies generally to any application-specific image enhancement algorithm. Published in IEEE PAMI (IF=5.07)