基于散射模型的水下浑浊图像增强方法  

Underwater turbidity image enhancement method based on scattering model

作　　者：张伟[1] 高赛博 李子轩 宫鹏 伍文华 ZHANG Wei;GAO Saibo;LI Zixuan;GONG Peng;WU Wenhua(Science and Technology on Underwater Vehicle Laboratory,Harbin Engineering University, Harbin 150001,China)

机构地区：[1]哈尔滨工程大学水下机器人技术重点实验室,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2021年第8期1209-1216,共8页Journal of Harbin Engineering University

基　　金：水下机器人国防重点实验室基金项目(JCKYS2020SXJQR-04).

摘　　要：由于水对光的选择性吸收和光的散射作用,水下图像存在色偏和雾化等现象,传统的去雾方法用于水下图像时效果欠佳。本文以散射模型为基础,提出了一种水下图像增强方法。基于局部区块熵分析估计背景光,权衡清晰度与图像逼真度设计了寻优透射率的代价函数,通过改进的加权最小二乘保边滤波算法细化透射率,设计了自适应调节因子对图像增强公式进行修正。实验结果表明:该算法能有效地去除水下图像的浑浊,提高图像的整体清晰度,验证了所提出算法工程有效性。Some underwater image phenomena,such as color deviation and atomization,are caused by the selective absorption of water and light scattering.The traditional dehazing method is not suitable for underwater images.A method for improving underwater images is proposed based on the scattering model.The background light estimation method based on local block entropy analysis is proposed,and the cost function for optimizing transmission is designed by weighing image edge gradient and fidelity,and refining transmission with an improved weighted least square edge-preserving filtering algorithm.The adaptive adjustment factor is then used to modify the image enhancement formula.Experimental results show that the proposed algorithm can effectively remove turbidity from underwater images,and improve the overall image clarity,proving the proposed algorithm′s engineering effectiveness.

关 键 词：水下视觉 浑浊图像 散射模型 图像增强 背景光 透射率 熵分析 最小二乘保边滤波 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]