自动化学报2025，Vol.51Issue(2)：376-390,15.DOI:10.16383/j.aas.c240491
基于颜色转移和自适应增益控制的混合水下图像增强
Hybrid Underwater Image Enhancement Based on Color Transfer and Adaptive Gain Control
摘要
Abstract
Aiming at handling the degradation problems of color deviation and low contrast in underwater images,a hybrid underwater image enhancement method based on color transfer and adaptive gain control is proposed.Firstly,the color transfer image and the maximum attenuation map-guided fusion strategy are used to correct the color deviation of underwater images.Secondly,the structure image of the V channel is obtained based on the first-order primal-dual algorithm to effectively suppress the noise interference.Furthermore,adaptive gain control based on the high-frequency information of the image is proposed to adaptively adjust the gain to obtain detailed image.Finally,the structure image and the detail image are weighted fused to obtain a high-quality underwater image.The experimental results demonstrate that for underwater images with various natural and industrial environments,1)The outputs of the proposed method can effectively correct the color distortion;2)It can improve the contrast of underwater images and suppress the noise interference;3)The quantitative metrics and high-level visual tasks(ob-ject detection,image segmentation,key-point detection,and underwater binocular distance measurement)are super-ior to other state-of-the-art underwater enhancement methods,which provides a useful reference for engineering ap-plications such as underwater automated grasping.关键词
水下图像增强/颜色校正/梯度平滑/纹理增强/水下成像模型Key words
Underwater image enhancement/color correction/gradient smoothing/texture enhancement/underwa-ter imaging model引用本文复制引用
崔峥,王森,王娴,段浩,杨春曦,那靖..基于颜色转移和自适应增益控制的混合水下图像增强[J].自动化学报,2025,51(2):376-390,15.基金项目
国家重点研发计划(2023YFE0204700),国家自然科学基金(62433012,62063011),云南省科技厅重大专项项目(202302AD 080005)资助Supported by National Key Research and Development Pro-gram of China(2023YFE0204700),National Natural Science Foundation of China(62433012,62063011),and Yunnan Major Scientific and Technological Projects(202302AD080005) （2023YFE0204700）
