Color correction and restoration based on multi-scale recursive network for underwater optical image

作者：

Highlights：

• A network is proposed to simulate underwater images based on style transfer method.

• We propose a dynamic threshold color correction method based on global information.

• A multi-scale recursive convolutional network is proposed to improve image clarity.

• The proposed method has obvious advantages in multiple underwater scenes.

摘要

•A network is proposed to simulate underwater images based on style transfer method.•We propose a dynamic threshold color correction method based on global information.•A multi-scale recursive convolutional network is proposed to improve image clarity.•The proposed method has obvious advantages in multiple underwater scenes.

论文关键词：Underwater image,Style transfer,Color correction,Image enhancement,Convolutional neural network

论文评审过程：Received 5 May 2020, Revised 23 August 2020, Accepted 14 January 2021, Available online 3 February 2021, Version of Record 11 February 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116174