《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (9): 2838-2844.DOI: 10.11772/j.issn.1001-9081.2021081433

• 多媒体计算与计算机仿真 • 上一篇    

基于多尺度网络的运动模糊图像复原算法

魏海云, 郑茜颖(), 俞金玲   

  1. 福州大学 物理与信息工程学院,福州 350108
  • 收稿日期:2021-08-12 修回日期:2021-11-20 接受日期:2021-11-25 发布日期:2022-01-07 出版日期:2022-09-10
  • 通讯作者: 郑茜颖
  • 作者简介:魏海云(1996—),女,河南信阳人,硕士研究生,主要研究方向:计算机视觉、图像处理、图像复原;
    俞金玲(1983—),女,福建福州人,教授,博士,主要研究方向:半导体以及拓扑绝缘体的自旋相关光电流。
  • 基金资助:
    国家自然科学基金资助项目(61471124);福建省科技重点产业引导项目(2020H0007)

Motion blurred image restoration algorithm based on multi-scale network

Haiyun WEI, Qianying ZHENG(), Jinling YU   

  1. College of Physics and Information Engineering,Fuzhou University,Fuzhou Fujian 350108,China
  • Received:2021-08-12 Revised:2021-11-20 Accepted:2021-11-25 Online:2022-01-07 Published:2022-09-10
  • Contact: Qianying ZHENG
  • About author:WEI Haiyun, born in 1996, M. S. candidate. Her research interests include computer vision, image processing, image restoration.
    YU Jinling, born in 1983, Ph. D., professor. Her research interests include spin-dependent photocurrents in semiconductors and topological insulators.
  • Supported by:
    National Natural Science Foundation of China(61471124);Fujian Provincial Key Science and Technology Industry Guidance Project(2020H0007)

摘要:

动态场景的非均匀盲去模糊一直是图像复原领域中的一个难题。针对目前的模糊图像复原算法不能很好地解决多样性模糊源的问题,提出了一种端到端的基于多尺度网络的运动模糊图像复原算法。所提算法使用修剪过的残差块作为基本单元,且在每一级尺度上都采用相同的非对称编解码网络。为了更好地提取输入图像特征,在编码端使用引入注意力机制的残差模块,还加入了空间金字塔池化层。编码端和解码端中间的循环单元可以获取图像的空间信息,从而利用图像空间的连续性来进行非均匀运动模糊图像的复原。测试结果显示,在GoPro数据集上所提算法的峰值信噪比(PSNR)达到33.69 dB,结构相似性(SSIM)达到0.953 7,且能够更好地复原模糊图像的细节信息,而在Blur数据集上所提算法的PSNR为31.47 dB,SSIM为0.904 7。实验结果表明,与尺度递归网络和深度层次化多patch网络相比,所提算法取得了更优的模糊图像复原效果。

关键词: 非均匀盲去模糊, 多尺度网络, 注意力机制, 残差模块, 循环单元

Abstract:

Non-uniform blind deblurring of dynamic scenes has always been a difficult problem in the field of image restoration. Aiming at the problem that the current blurred image restoration algorithms cannot solve the problem of diverse fuzzy sources well, an end-to-end motion blurred image restoration algorithm based on multi-scale network was proposed. In the proposed algorithm, the pruned residual blocks were used as the basic units, and the same asymmetric encoder-decoder network was used at each scale. In order to extract the features of the input image better, a residual module with attention mechanism was used in the coding side, and a spatial pyramid pooling layer was added. The recurrent unit between the encoding side and decoding side was able to obtain spatial information of the image, so that the image space continuity was able to be used to restore non-uniform motion blurred image. Test results show that the Peak Signal-to-Noise Ratio (PSNR) and Structural SIMilarity (SSIM) of the proposed algorithm are 33.69 dB and 0.953 7 respectively on GoPro dataset, and the blur image details can be recovered better, and the PSNR and SSIM of the proposed algorithm on Blur dataset are 31.47 dB and 0.904 7 respectively. Experimental results show that compared with scale-recurrent network and deep stacked hierarchical multi-patch network, the proposed algorithm achieves better blurred image restoration.

Key words: non-uniform blind deblurring, multi-scale network, attention mechanism, residual module, recurrent unit

中图分类号: