摘要
针对水下光衰减和悬浮粒子导致的水下图像蓝绿色偏和雾化问题,提出了一种基于颜色补偿和颜色线模型的水下图像增强算法。首先,通过改进颜色补偿算法对图像的衰减通道进行判断和补偿,以降低图像色偏程度并提高颜色线的拟合效果。其次,结合信息量评价指标和迭代阈值策略实现图像背景分离,以避免前景明亮物体的影响,从而提高背景光估计的准确性。然后,利用颜色线与背景光矢量构建了一种能够准确估计透射率的加速凸优化算法,以提高图像的去雾效果。最后,通过对颜色通道进行归一化处理得到没有色偏的无雾图像。增强后图像的平均峰值信噪比(PSNR)、水下彩色图像质量评价(UCIQE)、水下图像质量评价(UIQM)和彩色投射因子K分别达到25.36 dB、0.64、5.02和0.02。实验结果显示,相比其他算法,本文算法能更有效地解决水下图像的颜色失真和雾化现象。
Aiming to address the issues of color distortion and haziness in underwater images caused by underwater light attenuation and suspended particles,a color compensation and color line model-based underwater image enhancement algorithm was proposed.Firstly,by improving the color compensation algorithm to assess and compensate for the attenuation channels in the image,the degree of color deviation was reduced and the fitting effect of the color line was improved.Secondly,by combining information entropy evaluation metrics and iterative threshold strategies,image background separation was achieved to avoid the influence of bright foreground objects,thereby enhancing the accuracy of background light estimation.Subsequently,an accelerated convex optimization algorithm was constructed using the color line and background light vectors to accurately estimate the transmittance,thus improving the image dehazing effect.Finally,normalization of the color channels could result in haze-free images without color deviation.The enhanced images could achieve an average peak signal-to-noise ratio(PSNR)of 25.36 dB,underwater color image quality evaluation(UCIQE)of 0.64,underwater image quality measure(UIQM)of 5.02,and color cast factor K of 0.02.Experimental results demonstrate that compared to other algorithms,this method can more effectively address color distortion and haziness in underwater images.
作者
梁秀满
赵治刚
刘振东
于海峰
姚欣哲
LIANG Xiuman;ZHAO Zhigang;LIU Zhendong;YU Haifeng;YAO Xinzhe(College of Electrical Engineering,North China University of Science and Technology,Tangshan 063210,China)
出处
《应用光学》
北大核心
2025年第5期1034-1043,共10页
Journal of Applied Optics
基金
河北省自然科学基金(D2024209006)
河北省教育厅科学研究项目(QN2024147)
华北理工大学研究生创新项目(2024S05,2024S08)。
关键词
颜色补偿
颜色线模型
水下图像增强
背景光估计
凸优化
color compensation
color-line model
underwater image enhancement
background light estimation
convex optimization
