基于复原结构与增强纹理融合的水下图像清晰化  

Underwater image sharpening based on fusion of restored structure and enhanced texture

作　　者：林森 迟凯晨[4] 唐延东 LIN Sen;CHI Kai-chen;TANG Yan-dong(College of Automation and Electrical Engineering,Shenyang Ligong University,Shenyang 110159,China;State Key Laboratory of Robotics,Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110016,China;Institutes for Robotics and Intelligent Manufacturing,Chinese Academy of Sciences,Shenyang 110016,China;College of Electronic and Information Engineering,Liaoning Technical University,Huludao 125105,China)

机构地区：[1]沈阳理工大学自动化与电气工程学院,沈阳110159 [2]中国科学院沈阳自动化研究所机器人学国家重点实验室,沈阳110016 [3]中国科学院机器人与智能制造创新研究院,沈阳110016 [4]辽宁工程技术大学电子与信息工程学院,辽宁葫芦岛125105

出　　处：《控制与决策》2022年第3期635-644,共10页Control and Decision

基　　金：国家自然科学基金项目(91648118,61473280,61991413);辽宁省重点研发计划项目(2019JH2/10100014)。

摘　　要：针对水体对光的吸收与散射作用,导致水下拍摄图像存在雾化现象、色彩失真等问题,提出一种基于复原结构与增强纹理融合的水下图像清晰化算法.首先,通过相对总变差模型将图像分解为结构层与纹理层;其次,基于背景光的高亮度与平坦特性及颜色信息计算背景光值,利用红色暗通道先验优化透射率,通过逆求解成像模型得到复原结构层;然后,提出将梯度平滑方法用于纹理层,在抑制噪声的同时有效增强纹理细节;最后,融合复原结构层与增强纹理层,得到清晰的水下图像.实验结果表明,所提出算法将复原与增强技术相结合,清晰化处理后的图像较好地去除了雾化现象,且色彩鲜明、细节丰富.相对于各比较算法,水下彩色图像质量评价指标提高16.09%,为水下机器人等工程实践提供了可行的参考.Aiming at the problems of fogging and color distortion in underwater images caused by the absorption and scattering of light by water, an underwater image sharpening based on fusion of restored structure and enhanced texture is proposed. Firstly, the image is decomposed into a structure layer and a texture layer by the relative total variation model.Secondly, the background light value is calculated based on the high brightness, flat characteristics and color information of the background light, the transmittance is optimized by using the red dark channel prior. The restored structure layer is obtained by inversely solving the imaging model. Then, a gradient smoothing method is proposed for the texture layer,which can effectively enhance the texture details while suppressing the noise. Finally, the restored structure layer and the enhanced texture layer are fused to obtain a clear underwater image. The experimental results show that the proposed algorithm combines the restoration and enhancement technology, the sharpened image can better remove the fogging phenomenon, and the color is bright and the details are rich. Compared with the comparison algorithms, the underwater color image quality evaluation index is improved by 16.09%, which provides a feasible reference for engineering practices such as underwater robots.

关 键 词：图像清晰化 水下光学成像模型 暗通道先验 梯度平滑 颜色校正 机器视觉 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]