基于改进暗通道的水下图像快速复原  

Rapid restoration of underwater image based on improved dark channel

作　　者：龚万齐 冯常[1] GONG Wan-qi;FENG Chang(Institute of Optics and Electronics,Chinese Academy of Sciences,Chengdu 610209,China;School of Optoelectronics,University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院光电技术研究所,四川成都610209 [2]中国科学院大学光电学院,北京100049

出　　处：《计算机工程与设计》2022年第7期2015-2021,共7页Computer Engineering and Design

摘　　要：为解决水下图像由于受到水体对光线的吸收和散射作用而产生的图像退化问题,提出一种改进的水下暗通道先验方法。分析水下图像退化因素,通过色彩空间转换矫正和平衡图像三通道颜色信息,通过小尺度图像估计透射率图和全局背景光,结合暗通道先验模型完成图像的快速复原。对比实验的主观和客观分析结果表明,所提方法可以有效还原色彩信息,恢复水下图像的细节,获得较好的显示效果,显著提高图像复原处理速度,具有良好的鲁棒性和优越性。To solve the problem of image degradation caused by the absorption and scattering of light by water,an improved underwater dark channel prior method was proposed.The degradation factors of the underwater image were analyzed,and the color information of the three channels of the image was corrected and balanced by color space conversion.The transmissibility map and global background light were estimated through the small scale image,and the image was quickly restored by combining the dark channel prior model.Through the subjective and objective analysis of the comparative experiment,it is verified that the color information can be effectively restored,the details of the underwater image can be restored,the better display effect can be obtained,and the processing speed of image restoration can be significantly improved.It has a good robustness and superiority.

关 键 词：水下图像复原 色彩空间转换 暗通道先验 小尺度估计 图像质量评价 

分 类 号：TN911.73[电子电信—通信与信息系统]