计算机工程 ›› 2019, Vol. 45 ›› Issue (1): 217-220.doi: 10.19678/j.issn.1000-3428.0049076

• 图形图像处理 • 上一篇    下一篇

超深卷积神经网络的图像超分辨率重建研究

连逸亚,吴小俊   

  1. 江南大学 物联网工程学院,江苏 无锡 214122
  • 收稿日期:2017-10-25 出版日期:2019-01-15 发布日期:2019-01-15
  • 作者简介:连逸亚(1993—),男,硕士研究生,主研方向为图像超分辨率重建;吴小俊,教授、博士。
  • 基金项目:

    国家自然科学基金(61672265)。

Research on Image Super-Resolution Reconstruction of Super Deep Convolutional Neural Network

LIAN Yiya,WU Xiaojun   

  1. College of Internet of Things Engineering,Jiangnan University,Wuxi,Jiangsu 214122,China
  • Received:2017-10-25 Online:2019-01-15 Published:2019-01-15

摘要:

针对VDCN网络结构在大尺度因子上超分辨率效果较差的缺点,提出一种高精度单图像超分辨率重建方法。将ReLU激活函数更换为PReLU激活函数,增加网络层数,使用25个带PReLU激活函数的卷积层进行训练和测试。实验结果表明,与VDCN方法相比,该方法耗费时间较少,且性能更稳定。

关键词: 卷积神经网络, 图像超分辨率, PReLU激活函数, 深度学习, 网络深度

Abstract:

To solve the disadvantage of VDCN network structure in large scale factors,a new high-precision single image super-resolution method is proposed.The ReLU activation function is replaced with the PReLU activation function,and the number of network layers is increased.The model uses 25 convolution layers with PReLU activation functions to train and test.Experimental results show compared with the VDCN method,this method takes less time and has stable performance.

Key words: Convolutional Neural Network(CNN), image Super-Resolution(SR), PReLU activation function, deep learning, network depth

中图分类号: