Computer Engineering and Applications ›› 2021, Vol. 57 ›› Issue (5): 168-172.DOI: 10.3778/j.issn.1002-8331.2003-0336

Previous Articles     Next Articles

Improved Image Denoising Generative Adversarial Network Algorithm

CHEN Renhe, LAI Zhenyi, QIAN Yurong   

  1. School of Software, Xinjiang University, Urumqi 830046, China
  • Online:2021-03-01 Published:2021-03-02

改进的生成对抗网络图像去噪算法

陈人和,赖振意,钱育蓉   

  1. 新疆大学 软件学院,乌鲁木齐 830046

Abstract:

The existence of image noise will disturb people’s understanding of the image. In order to effectively remove the noise and obtain a better visual perception, an algorithm based on generative adversarial network is proposed. This algorithm obtains more image features by increasing the width of the generative network. And it adds a global residual to the feature extraction and learning of the input noise image to avoid feature loss. The network uses a weighted sum of the anti-loss and reconstruction loss, which can effectively retain the details of the image while removing noise. Experimental results show that the algorithm can effectively remove image noise and improve the visual perception of the image.

Key words: generative adversarial network, image denoising, global residual, reconstruction loss

摘要:

由于图像噪声的存在会干扰人对图像的理解,为了有效地去除噪声并获得比较好的视觉观感,提出一种基于生成对抗网络算法,该算法通过增加生成网络的宽度来获取更多的图像特征,并加入一个全局残差对输入的噪声图像进行特征的提取与学习,避免特征的丢失。网络采用对抗损失和重建损失的加权和,在去除噪声的同时能够有效地保留图像的细节信息。实验结果表明,该算法能够有效地去除图像噪声,改善图像的视觉观感。

关键词: 生成对抗网络, 图像去噪, 全局残差, 重建损失