Research on image enhancement algorithms for deep-sea hydrothermal vent areas based on deep learning

In this paper, a deep-sea hydrothermal area image enhancement method is proposed based on the conditions of manned deep submarine active illumination system. First, according to the characteristics of deep-sea images, red channel compensation and grey scale world are used to correct the colour bias of deep-sea images. Then, the brightness and contrast of the image are adjusted globally and locally by combining the active illumination characteristics, the enhanced gamma correction and the CLAHE algorithm are used to obtain the enhanced image, respectively, the detailed features of the image are extracted in depth by using the VGG19 network, and the image is fused at multiple scales by combining the adaptive weighted fusion and the deep convolution, and finally, a deep-sea hydrothermal area image with the restored colours is obtained. In this paper, the proposed method can be better adapted to the characteristics of deep-sea hydrothermal area images and achieve better results by combining the methods and using the observation images of deep-sea hydrothermal area of a voyage of "Jiaolong" for verification.