杨爱萍¹, 田鑫¹, 杨炳旺¹, 王建^1,2
AuthorsHTML:杨爱萍¹, 田鑫¹, 杨炳旺¹, 王建^1,2
AuthorsListE:Yang Aiping¹, Tian Xin¹, Yang Bingwang¹, Wang Jian^1,2
AuthorsHTMLE:Yang Aiping¹, Tian Xin¹, Yang Bingwang¹, Wang Jian^1,2
Unit:1. 天津大学电气自动化与信息工程学院，天津 300072；2. 国家海洋技术中心，天津 300112
Unit_EngLish:1．School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China
2．National Ocean Technology Center, Tianjin 300112, China
Abstract_Chinese:水体对光有较强的吸收和散射作用, 从而造成水下图像颜色失真、对比度下降和细节模糊等问题．针对上述问题, 基于图像融合框架提出一种水下图像清晰化方案．首先, 结合水下成像模型, 提出了基于景深估计的颜色校正方法, 其可以有效去除水下图像的蓝(绿)色基调, 同时有效解决红通道过补偿问题．然后, 对水下图像进行颜色校正和对比度增强得到两幅输入图像; 最后, 结合亮度、对比度、饱和度、显著性等特征构造权重图将两输入图进行多尺度融合, 得到最后的复原图像．实验结果表明, 该算法可以有效地去除水下图像颜色失真, 同时能够大幅提升图像的对比度和清晰度．
Abstract_English:Due to the absorption and scattering effect on light when traveling in water，there are three major problems of underwater imaging: low contrast，color distortion and fuzzy image details. In view of the above problems，a scheme for sharpening underwater images has been proposed based on image fusion framework. First of all，combined with the underwater imaging model，a color correction method based on depth estimation was proposed which can effectively remove the blue(green)color tone of the underwater image and solve the problem of red channel overcompensation. Then，two input images were obtained by color correction and contrast enhancement of the underwater image. Finally，the restored image was obtained by utilizing the multi-scale fusion strategy which is based on the weighted maps constructed by combining the features of brightness，contrast，saturation and saliency. The experimental results show that this algorithm can effectively remove the color distortion of underwater images and greatly improve the contrast and sharpness of images.
Keyword_Chinese:多尺度融合; 颜色校正; 对比度增强; 权重图
Keywords_English:multi-scale fusion; color correction; contrast enhancement; weighted map

PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=6092