Residual Dense Network for Image Super-Resolution

Abstract

A very deep convolutional neural network (CNN) has recently achieved great success for image super-resolution (SR) and offered hierarchical features as well. However, most deep CNN based SR models do not make full use of the hierarchical features from the original low-resolution (LR) images, thereby achieving relatively-low performance. In this paper, we propose a novel residual dense network (RDN) to address this problem in image SR. We fully exploit the hierarchical features from all the convolutional layers. Specifically, we propose residual dense block (RDB) to extract abundant local features via dense connected convolutional layers. RDB further allows direct connections from the state of preceding RDB to all the layers of current RDB, leading to a contiguous memory (CM) mechanism. Local feature fusion in RDB is then used to adaptively learn more effective features from preceding and current local features and stabilizes the training of wider network. After fully obtaining dense local features, we use global feature fusion to jointly and adaptively learn global hierarchical features in a holistic way. Experiments on benchmark datasets with different degradation models show that our RDN achieves favorable performance against state-of-the-art methods.

Adaptive Densely Connected Single Image Super-Resolution

“…In order to verify the validity of the model , we compare the performance on five standard benchmark datasets: Set5 [1], Set14 [27], B100 [17], Urban100 [8], and manga109 [18]. In terms of PSNR, SSIM and visual effects, We compare our models with the state-of-theart methods including Bicubic, SRCNN [5], VDSR [10], LapSRN [12], MemNet [22], EDSR [16], RDN [32], RCAN [31], SAN [4]. We also adopt self-ensemble strategy [16] to further improve our ADCSR and denote the selfensembled ADCSR as ADCSR+.…”

“…Lim et al proposed an enhanced depth residual network (EDSR) [16], which made a significant performance through the deeper network. Other deep network like RDN [32] and Mem-Net [22], are based on dense blocks. Some networks focus on feature correlations in channel dimension, such as RCAN [31]and SAN [4].…”

“…This problem was solved when He et al [6] proposed residual net (ResNet), and Huang et al [7] proposed dense net (DesNet). Many large scale networks were introduced in SISR, such as SRResNet [14], EDSR [16], SR-DenseNet [24], RDN [32] etc.These methods aim at building a deeper network to increase the performance. Other methods such as RCAN [31] and SAN [4] try to learn the correlation of the features in the middle layers.…”

Example‐Based Colourization Via Dense Encoding Pyramids

“…Thus, taking full advantage of multi‐scale features is more important than learning more redundant features. In our paper, we propose a parallel structure based on RDB [ZTK*18] to extract low‐ and high‐level information effectively in a wider network. The residual and dense structure allows the network to learn low‐level information and extracts features than the normal convolution.…”

FHDR: HDR Image Reconstruction from a Single LDR Image using Feedback Network

“…The first block is the Feature Extraction block (FEB), followed by the Feedback block (FBB) and an HDR reconstruction block (HRB). Inspired by [10], we use a global residual skip connection for bypassing low level LDR features at every iteration to guide the HDR reconstruction block in the final layers. For every training example, the network runs for n iterations.…”