Multi-input fusion adversarial network for underwater image enhancement

被引：0
|
作者：
Lin S. [1 ,2 ,3 ]
Liu S. [1 ]
Tang Y. [2 ,3 ]
机构：
[1] Electronic and Information Engineering School, Liaoning Technical University, Huludao
[2] State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang
[3] Institutes for Robotics and Intelligent Manufacturing, Chinese Academy of Sciences, Shenyang
来源：
Hongwai yu Jiguang Gongcheng/Infrared and Laser Engineering | 2020年 / 49卷 / 05期
关键词：
Deep learning; Encoding and decoding structure; Generative adversarial network; Multi-input fusion; Underwater image enhancement;
D O I：
10.3788/IRLA20200015
中图分类号：
学科分类号：
摘要：
For underwater image of low contrast, color deviation and blurred details and other issues, the multi-input fusion adversarial networks was proposed to enhance underwater images. The main feature of this method was that the generative network used encoding and decoding structure, filtering noise through convolution layer, recovering lost details through deconvolution layer and refining the image pixel by pixel. Firstly, the original image was preprocessed to obtain two types of images: color correction and contrast enhancement. Secondly, the confidence graph of the difference between the two enhanced images and the original image was learned by using the generated network. Then, in order to reduce artifacts and details blur introduced by the two enhancement algorithms in the process of generating network learning, the texture extraction unit was added to extract texture features from the two enhanced images, and the extracted texture features were fused with the corresponding confidence map. Finally, the enhanced underwater image was obtained by constructing multiple loss functions and training the adversarial network repeatedly. The experimental results show that the enhanced underwater image has bright color and improved contrast, the average value of UCIQE and NIQE is 0.639 9 and 3.727 3 respectively. Compared with other algorithms, the algorithm has significant advantages and proves its good effect. © 2020, Editorial Board of Journal of Infrared and Laser Engineering. All right reserved.
引用
收藏
相关论文
共 15 条
  • [1] Hu H., Li X., Liu T., The latest development of underwater image restoration technology based on polarization imaging, Infrared and Laser Engineering, 48, 6, (2019)
  • [2] Chen Q., Zhang X., Application of hybrid residual learning and guided filtering algorithm in image defogging, Optical Precision Engineering, 27, 12, pp. 2702-2712, (2019)
  • [3] Fabbri C., Islam M.J., Sattar J., Enhancing underwater imagery using generative adversarial networks, IEEE International Conference on Robotics and Automation(ICRA), pp. 7159-7165, (2018)
  • [4] Guo Y., Li H., Zhuang P., Under-water image enhancement using a multiscale dense generative adversarial network, IEEE Journal of Oceanic Engineering, (2019)
  • [5] Li C., Wand M., Precomputed real-time texture synthesis with markovian generative adversarial networks, European Conference on Computer Vision Springer, pp. 702-716, (2016)
  • [6] Ye X., Xu H., Ji M., Et al., Underwater image enhancement using stacked generative adversarial networks, Pacific Rim Conference on Multimedia, pp. 514-524, (2018)
  • [7] Liang T., Zhang X., Duan P., Et al., Target detection of underwater strong scattering medium by improved dark channel method, Infrared and Laser Engineering, 49, 2, (2020)
  • [8] He K., Sun J., Fellow, Et al., Single image haze removal using dark channel prior, IEEE Transactions on Pattern Analysis & Machine Intelligence, 33, 12, pp. 2341-2353, (2011)
  • [9] Wang C., An J., Jiang X., Et al., Candidate region optimization algorithm based on convolutional neural network, Chinese Optics, 12, 6, pp. 1348-1361, (2019)
  • [10] Jonathan L., Evan S., Trevor D., Fully convolutional networks for semantic segmentation, IEEE Conference on Computer Vision and Pattern Recognition(CVPR), pp. 3434-3440, (2015)
12