Simultaneous single image super‐resolution and blind Gaussian denoising via slim ghost full‐frequency residual blocks

Saghar Farhangfar,Aryaz Baradarani,Mohammad Asadpour,Mohammad Ali Balafar,Roman Gr. Maev

DOI: https://doi.org/10.1049/ipr2.13230

IF: 2.3

2024-11-01

IET Image Processing

Abstract:This paper introduces a novel generative adversarial network‐based model designed for simultaneous super‐resolution and blind additive white Gaussian noise denoising. The model comprises two key components: netdeg for blind denoising and netSR for super‐resolution. netdeg restores a clean low‐resolution image using innovative ghost residual blocks and a frequency separation module, while netSR utilizes slim ghost full‐frequency residual blocks for processing both low and high‐frequency information. Experimental results demonstrate that the proposed model outperforms existing methods in handling additive white Gaussian noise at various levels, offering robust and superior performance without requiring specific setup conditions. Given that super‐resolution (SR) aims to recover lost information, and low‐resolution (LR) images in real‐world conditions might be corrupted with multiple degradations, considering basic bicubic down‐sampling as the sole degradation significantly limits the performance of most existing SR models. This paper presents a model for simultaneous super‐resolution and blind additive white Gaussian noise (AWGN) denoising with two components (netdeg and netSR) that is based on a generative adversarial network (GAN) to achieve detailed results. netdeg, featuring residual and innovative cost‐effective ghost residual blocks with a frequency separation module for obtaining long‐range information, blindly restores a clean version of the LR image. netSR leverages slim ghost full‐frequency residual blocks to process low‐frequency (LF) and high‐frequency (HF) information via static large convolutions and pixel‐wise highlighted input‐adaptive dynamic convolutions, respectively. To address the susceptibility of dynamic layers to noise and preserve feature diversity while reducing model's costs, static and dynamic layer features are combined and highlighted. Diverse and non‐redundant features are then processed using ghost‐style blocks. The proposed model achieves comparable SR results in bicubic down‐sampling scenarios, outperform existing SR methods in the complex task of concurrent SR and AWGN denoising, and demonstrate robustness in handling images corrupted with varying levels of AWGN.

computer science, artificial intelligence,engineering, electrical & electronic,imaging science & photographic technology