Deep Differential Convolutional Network for Single Image Super-Resolution

Abstract

The deep convolutional neural networks and residual networks have shown great success and high-quality reconstruction for single image super-resolution. It is clearly seen that among the bestknown super-resolution models, deep learning-based methods demonstrate state-of-the-art performance. In this paper, we propose a deep differential convolutional network (DCN) for single image superresolution (SRDCN). The proposed DCN is a novel convolutional network, which is composed of convolutional layers, parametric rectified linear units (PReLU), and the identity skip connection. Different from other deep learning-based methods which complete the reconstruction by learning the mapping function between low-resolution and high-resolution images, the proposed algorithm makes changes to the way of reconstruction. In the proposed network, we use DCN to obtain the reconstructed images and the differences between the low-resolution and reconstructed images in the reconstruction process. Then the differences combined with the original low-resolution image and the reconstructed image that from the last DCN are used for final reconstruction. In addition, the loss function is more rationally designed and optimized in this paper. The proposed loss function contains three parts of loss: feature loss, style loss, and mean squared error (MSE) loss. These losses will be used to supervise the structure and content of the reconstructed image. The experimental results prove that the proposed model is superior to many state-of-the-art super-resolution methods in terms of both peak signal-to-noise ratio (PSNR) and structural similarity index metrics (SSIM).INDEX TERMS Deep convolutional neural networks, differential convolutional network, single image super-resolution, peak signal-to-noise ratio, structure similarity index metrics.

Deep Learning-based Image Super-resolution Algorithms - A Survey

“…Liu et al 2019 proposed a SR using deep convolution network (SRDCN) which consists of parametric-rectified linear units, skip connection of identity, and convolutional layers [12]. It combines the differences between LR and the restricted image for getting final SR image.…”

“…Further it mentions the merits and demerits of different SR schemes. This work analyses recently developed NN models such as fast conceptual deep auto encoder (FCDA) [8], larger dictionary model (LDM) [9], spatial transform to reduce geometrical effect (STRGE) [10], low-complexity convolution kernel (LCCK) [11], deep convolution with residual network (DCRN) [12], recurrent fusion network (RFN) [13] , generative adversarial network and multi-perspective discriminator (GAN & MPD) [14], component learning (CL) [15], deep and shallow network (DSN) [16], complementary priors and ensemble learning (CPEL) [17] and two discriminator network (TDN) [18]. So, the major contributions of this are: briefing different deep learning architecture recently used for image super resolution, comparing the characteristics of different architecture in terms of NN model and quality, and suggesting research scope for future developments.…”

Lightweight Image Super-Resolution With Mobile Share-Source Network

“…As the HUNG et al came up with SSNet-M, he proposed an image super-resolution network with extremely small parameters and operations, with the recursive blocks being used flexibly [15], [16]. The Deep Differential Convolutional network has been proposed by Peng tried to modify the loss function [23]. Also, the addition of the Wavelets and color guidance block was impressive [1].…”

Split-Attention Multiframe Alignment Network for Image Restoration

“…Image restoration, the problem of restoring a clean image from its degraded version, which includes subtasks such as image denoising [1], image superresolution [2], image decompression [3] and image deblurring [4], is an ill-posed problem. Because multiple consecutive images can provide complementary information, especially in distorted areas, occluded areas and motion-blurred areas in images, utilizing multiple images provides a new way to improve the accuracy of image restoration algorithms.…”