提升深度残差网络：单图像超分辨率研究

需积分: 16 68 浏览量更新于2024-08-26 收藏 5.5MB PDF 举报

"《增强深度残差网络在单图像超分辨率中的应用》(Enhanced Deep Residual Networks for Single Image Super-Resolution)是一篇由Bee Lim、Sanghyun Son、Heewon Kim、Seungjun Nah和Kyoung Mu Lee等来自首尔国立大学电子工程系及先进科学技术研究所的研究人员合作撰写的论文。随着深度卷积神经网络（DCNN）的发展，超分辨率技术的研究取得了显著进步，特别是通过残差学习方法，其性能得到了提升。该论文的焦点在于提出了一种新型的增强深度超分辨率网络(EDSR)，它超越了当前最先进的超分辨率方法。作者的关键贡献在于对传统残差网络进行了优化，移除了不必要的模块，这显著提高了模型的性能。他们不仅注重模型的精度，还关注训练过程的稳定性，通过扩大模型规模实现了进一步的性能提升。此外，研究人员提出了一个多尺度深度超分辨率系统(MDSR)和一种新的训练方法。这一创新允许在一个单一模型内同时处理不同放大倍数的高分辨率图像重建，从而节省了资源并提升了效率。这种方法证明了在处理不同尺度的图像超分辨率任务时，既能保持高质量的输出，又能实现高效的一体化解决方案。这篇论文展示了如何通过简化和扩展深度残差网络结构，结合多尺度策略，有效地解决单图像超分辨率问题，并且在实际应用中取得了卓越的性能。这为后续的超分辨率研究提供了有价值的新视角和技术参考，对于提高图像恢复的清晰度和细节层次具有重要意义。"

Enhanced Deep Residual Networks for Single Image Super-Resolution

Bee Lim Sanghyun Son Heewon Kim Seungjun Nah Kyoung Mu Lee

Department of ECE, ASRI, Seoul National University, 08826, Seoul, Korea

forestrainee@gmail.com, thstkdgus35@snu.ac.kr, ghimhw@gmail.com

seungjun.nah@gmail.com, kyoungmu@snu.ac.kr

Abstract

Recent research on super-resolution has progressed with

the development of deep convolutional neural networks

(DCNN). In particular, residual learning techniques exhibit

improved performance. In this paper, we develop an en-

hanced deep super-resolution network (EDSR) with perfor-

mance exceeding those of current state-of-the-art SR meth-

ods. The signiﬁcant performance improvement of our model

is due to optimization by removing unnecessary modules in

conventional residual networks. The performance is further

improved by expanding the model size while we stabilize

the training procedure. We also propose a new multi-scale

deep super-resolution system (MDSR) and training method,

which can reconstruct high-resolution images of different

upscaling factors in a single model. The proposed methods

show superior performance over the state-of-the-art meth-

ods on benchmark datasets and prove its excellence by win-

ning the NTIRE2017 Super-Resolution Challenge [26].

1. Introduction

Image super-resolution (SR) problem, particularly sin-

gle image super-resolution (SISR), has gained increasing

research attention for decades. SISR aims to reconstruct

a high-resolution image I

from a single low-resolution

image I

. Generally, the relationship between I

and

the original high-resolution image I

can vary depending

on the situation. Many studies assume that I

is a bicubic

downsampled version of I

, but other degrading factors

such as blur, decimation, or noise can also be considered for

practical applications.

Recently, deep neural networks [11, 12, 14] provide sig-

niﬁcantly improved performance in terms of peak signal-to-

noise ratio (PSNR) in the SR problem. However, such net-

works exhibit limitations in terms of architecture optimality.

First, the reconstruction performance of the neural network

models is sensitive to minor architectural changes. Also, the

same model achieves different levels of performance by dif-

0853 from DIV2K [26]

(PSNR / SSIM)

Bicubic

(30.80 dB / 0.9537)

VDSR [11]

(32.82 dB / 0.9623)

SRResNet [14]

(34.00 dB / 0.9679)

EDSR+ (Ours)

(34.78 dB / 0.9708)

Figure 1: ×4 Super-resolution result of our single-scale SR

method (EDSR) compared with existing algorithms.

ferent initialization and training techniques. Thus, carefully

designed model architecture and sophisticated optimization

methods are essential in training the neural networks.

Second, most existing SR algorithms treat super-

resolution of different scale factors as independent prob-

lems without considering and utilizing mutual relationships

among different scales in SR. As such, those algorithms re-

quire many scale-speciﬁc networks that need to to be trained

independently to deal with various scales. Exceptionally,

arXiv:1707.02921v1 [cs.CV] 10 Jul 2017

下载后可阅读完整内容，剩余8页未读，立即下载

路漫漫其修远.

粉丝: 3570
资源: 33

提升深度残差网络：单图像超分辨率研究

EDSR(torch)-Enhanced Deep Residual Networks for Single Image Super-Resolution代码

EDSR-Enhanced Deep Residual Networks for Single Image Super-Resolution代码

Enhanced Deep Residual Networks for Single Image Super-Resolution code

super_resolution.zip

A Comprehensive Review of Deep Learning-based Single Image S

基于深度学习的红外图像超分辨率重建.pdf

一个基于ncnn推理框架和qt搭建的一个super resolution的Demo, .zip

Python-EnhancedDeepResidualNetworksforSingleImageSuperResolution论文的PyTorch实现

【Project Practicality】: New Horizons in Image Transformation: A Practical Guide to the Application...

residual scaling layers

最新资源