Computer Engineering and Applications ›› 2020, Vol. 56 ›› Issue (8): 185-191.DOI: 10.3778/j.issn.1002-8331.1901-0331

Previous Articles     Next Articles

Image Super-Resolution Reconstruction Algorithm Based on Residual Network

CHEN Chen, LIU Mingming, LIU Bing, ZHOU Yong   

  1. 1. College of Computer  Science and Technology, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
    2. Institute of Intelligent Manufacturing, Jiangsu Vocational Institute of Architectural Technology, Xuzhou, Jiangsu 221008, China
    3. Aerospace Information Research Institute, Chinese Academy of Sciences, Xuzhou, Jiangsu 221116, China
  • Online:2020-04-15 Published:2020-04-14

基于残差网络的图像超分辨率重建算法

陈晨,刘明明,刘兵,周勇   

  1. 1.中国矿业大学 计算机科学与技术学院,江苏 徐州 221116
    2.江苏建筑职业技术学院 智能制造学院,江苏 徐州 221008
    3.中国科学院 航空航天信息研究所,江苏 徐州 221116

Abstract:

The traditional convolutional neural network uses end-to-end mapping between high/low resolution images based on sparsely expressed super-resolution images. The input image is high resolution and the output image is low resolution. SRCNN with three convolution layers has a certain reconstruction effect, but the field of receptivity is low. Therefore, a deeper network structure is proposed. This improvement makes the latter network layer have a larger field of perception, so that the result pixels can be inferred from more pixels. However, considering the influence of network structure deepening on transmission rate, the learning rate is improved by introducing a combination of local residual learning and global residual learning. The convergence rate is effectively accelerated by this method. The experimental results are verified. Compared with existing Bicubic, SRCNN and VDSR, the reconstruction effect has been improved in peak signal-to-noise ratio, structural similarity and visual effect.

Key words: convolutional neural network, image super-resolution reconstruction, residual network

摘要:

传统的卷积神经网络用到的方法是在稀疏表示的超分辨率图像的基础上学习高/低分辨率图像之间端到端的映射,输入的是高分辨率的图像,输出的是低分辨率的图像,拥有三层卷积层的SRCNN虽然有一定的重建效果,但是感受野较低,因此,提出加深网络结构的方法,此次改进使得后面的网络层拥有更大的感受野,这样结果的像素点可以根据更多的像素点来推断。但是考虑到网络结构加深对传输速率的影响,通过引入局部残差学习和全局残差学习相结合的方法来提高学习率,通过该办法有效地加快了收敛速度,并且通过实验结果验证,与已有的Bicubic、SRCNN和VDSR相比,重建效果在峰值信噪比、结构相似性和视觉效果上均有所提升。

关键词: 卷积神经网络, 图像超分辨率重建, 残差网络