The paper "Two-step domain adaptation for underwater image enhancement" supervised by Prof. Yunfeng Zhang inSchool of ComputerScience and Technology and written byaMaster's degree student Qun Jiang as the first author is now published online in the journalPattern Recognition.
The paper "Two-step domain adaptation for underwater image enhancement" supervised by Prof. Yunfeng Zhang in School of Computer Science and Technology and written by student Qun Jiang as the first author is now published online in the journalPattern Recognition.This journal is one of the top journals in the field of pattern recognition,and isJCR Zone 1 (Q1) Top Journal, with the latest impact factor of 7.74, which is highly influential in the field of machine learning and pattern recognition in the worldranked in Class A1 journals of our university.
Underwater images are the most effective tools for exploring the marine environment, but their practical applications are seriously affected by the degradation of underwater images due to absorption and scatteringofwavelength-dependent light in water. Underwater image enhancement is a classical underwater image processing method that aims to remove the effects of light scattering (similar to fog) and correct color aberrations. In this thesis, migration learning is applied to underwater image enhancement for the first time, and a new two-step domain adaptation framework for underwater image enhancement in real scenarios is proposed to migrate airborne image defogging to underwater image enhancement. And the method does not need to be trained with synthetic pairs of underwater images,thuseliminating the dependence on underwater truthful images.
The algorithm can be applied to subsequent practical application scenarios such as underwater target detection, marine life monitoring, ocean exploration, and underwater cable detection.
Editor: Li Yunzhi
