Single image super-resolution in compressed domain based on field of expert prior
Jia Xiao, Chen Wang, Xiyuan Hu · 2012
Considerable attention has been directed to the problem of producing high-resolution image from multiple or single compressed low-resolution images. Traditional superresolution (SR) reconstruction techniques that have been designed for raw (i.e., uncompressed) images may not be effective when applied to compressed images because they do not incorporate the compression process into their models. In this paper, a new SR method for compressed images is proposed. The distortion caused by compression is modeled as additive, spatially correlated Gaussian noise, while the original image is modeled as a high order Markov random field (MRF) based on the recently proposed Fields of Experts (FoE) framework. To further restore the strong edge, total variation (TV) image prior is also adopted in these areas. The whole algorithm achieve high detail preservation and high PSNR.