System and method for content-adaptive super-resolution via cross-scale self-learning

摘要

A display system converts Full-High Definition (FHD) image content into Ultra-High Definition (UHD) image content. The display system includes an interface that provides image content to a display having a first resolution. The display system also includes a processor that converts an input image from a second resolution to the first resolution. The first resolution is greater than the second resolution. The processor generates a first value corresponding to a lost high-frequency component (HFC) of the input image; generates an initial component of an output image; selects a patch of the input image; estimates a lost HFC of the patch of the initial component based on a portion of the lost HFC of the input image that corresponds to the selected patch; and generates the output image based on a sum of the patch of the initial component and the estimated lost HFC of the patch of the initial component.