Marine Snow Removal
Dongsheng Guo, Yiqing Huang, Tianshun Han, Haiyong Zheng, Zhaorui Gu, Bing Zheng · OCEANS 2022 - Chennai · 2022
High-quality underwater images are significant for marine-exploration-related works. However, capturing high-quality images in the water medium is extremely difficult. Marine snow, consisting of various organic and inorganic materials, is one of the degradation sources. However, there are no available marine snow dataset to further research the algorithm of marine snow removal problem. This paper analyse the characteristics of marine snow from the observations of researches and real underwater images and introduces a new dataset for marine snow removal of both paired and unpaired underwater image pairs. We then treat the marine snow removal task as an image-to-image translation problem and provide benchmark experiments on our dataset for references.