Underwater image enhancement algorithm based on multi-branch enhancement and fusion attention mechanism
Due to the refraction and absorption of light by water,underwater images are often severely degraded,such as color deviation,blurring and low visibility.In order to improve the visibility of underwater images,an underwater image enhancement network MBFA-GAN based on multi-branch enhancement and fusion attention mechanism is proposed.Firstly,by analyzing the color degradation and blur factors of underwater images,a color temperature repair module and a blur recovery module are designed for color correction and blur recovery of underwater images.Then,considering the complementarity of multiple branch features,the cyclic merging strategy is used to fuse the features enhanced by multiple branches with adap-tive fusion module to gradually enhance the image details.Finally,a fusion attention module is designed to deeply mine the correlation matrix in channel dimension and pixel dimension to improve the authenticity of enhanced images.The experimental results show that compared with the existing algorithms,the proposed underwater image enhancement algorithm has bet-ter deblurring effect and more real color,and can effectively improve the problem of underwa-ter image color deviation and blurring.
万方数据
维普
