Research and Implementation of Underwater Image Super-resolution Reconstruction Algorithm

曹虎（1997-），男，主要研究方向：物联网技术

段鹏瑞（1974-），男，副教授，主要研究方向：物联网技术

Abstract：The ocean occupies about three-quarters of the earth\'s surface area and contains abundant resources. Underwater images are an important information carrier for the development of marine resources. However, due to the scattering and absorption of seawater, underwater images have problems of insufficient contrast, color attenuation and blurring. In this paper, a super-resolution reconstruction method based on deep learning is used to restore high-quality underwater images. Aiming at the problem of insufficient contrast, this paper uses a salient object detection method to determine the saliency boundary of the object with the help of channel attention and spatial attention mechanisms, and calculates the average pixel intensity between the object and the background as the contrast loss. Aiming at the problem of color attenuation, a color loss based on color matrix is proposed to improve the performance of the model in color restoration. The experimental results show that the method proposed in this paper has obvious effects on contrast and color restoration in underwater image super-resolution reconstruction, and performs better in visual effects and objective indicators.

Keywords： computer application technology image super resolution generative adversarial networks salient object detection