Chinese Journal of Liquid Crystals and Displays, Volume. 37, Issue 6, 768(2022)

Underwater image enhancement method based on improved conditions generate adversarial networks

Yan-fei PENG, Jian LI*, Li-rui GU, and Man-ting ZHANG
Author Affiliations
  • School of Electronic and Information Engineering,Liaoning Technical University,Huludao 125105,China
  • show less

    The propagation of light in water severely uates red light, this leads to green or blue color difference of the underwater image. Against this red light attenuation phenomenon, a method of improved conditional generation adversarial network is proposed to enhance the underwater images. Raw images are firstly corrected for preliminary color using a dynamic threshold. Then, the color recovery of the underwater image is achieved by generating confrontation network using the condition and introducing the link block to learn the mapping relationship between the underwater image and the normal image. The image denoising is performed using a two-sided filtering algorithm to improve image visibility. The combination of L1 with L2 loss is introduced to learn image color, and the focus loss is added to solve the high imbalance of the sample proportion. This method has an excellent enhancement in both color distortion and blur of underwater images, and good visual effects are obtained.

    Tools

    Get Citation

    Copy Citation Text

    Yan-fei PENG, Jian LI, Li-rui GU, Man-ting ZHANG. Underwater image enhancement method based on improved conditions generate adversarial networks[J]. Chinese Journal of Liquid Crystals and Displays, 2022, 37(6): 768

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category:

    Received: Dec. 17, 2021

    Accepted: --

    Published Online: Jun. 22, 2022

    The Author Email: Jian LI (1044946077@qq. com)

    DOI:10.37188/CJLCD.2021-0327

    Topics