Learning Approaches for Super-Resolution Imaging
Wan-Chi Siu, Zhi-Song Liu, Junjie Huang, Kwok-Wai Hung · 2018
Image super-resolution is a topic of great interest. It has significantapplications in ultra-high-definition TV display, image resizing, facerecognition, object recognition, video coding and surveillance. Theobjective is to construct a high-resolution image from one orseveral low-resolution images, while minimizing visual artifacts.Classical approaches have come to a quality limit because of theconstraint on manual filter design and limited design structures.Learning approaches allow super-resolution algorithm design tobe adaptive to training data and automatically form thousands offilters/adapters for the best super-resolution. In this chapter, wewillintroduce (i) some conventional learning approaches, (ii) randomforests, and (iii) Convolutional Neural Network (CNN) for effectiveimage super-resolution.