基于单幅雾天图像的改进迭代白平衡算法  

Improved iterative white balance algorithm based on a single foggy image

作　　者：高智慧 翟艺书[1] 刘月圆 GAO Zhihui;ZHAI Yishu;LIU Yueyuan(School of Science,Tianjin University of Technology and Education,Tianjin 300222,China)

机构地区：[1]天津职业技术师范大学理学院,天津300222

出　　处：《高师理科学刊》2023年第11期21-26,共6页Journal of Science of Teachers＇College and University

基　　金：天津市高等学校科学技术发展基金项目(JWK1603)。

摘　　要：针对雾天环境下户外系统拍摄的图像呈现颜色失真、饱和度低、整体偏雾色表现等问题,在几种白平衡算法基础上,提出一种改进的迭代白平衡算法.对图像进行直方图拉伸变换扩展动态范围,将图像从RGB空间转化到YCbCr空间,根据亮度值特征筛选白色参考点进行颜色校正,以输入图像和输出图像直方图的余弦相似度作为终止条件,完成对白平衡迭代处理.实验结果表明,该算法可以有效地改善雾天图像色偏现象和颜色视觉效果.Aiming at the problems such as color distortion,low saturation and overall foggy color performance of images taken by outdoor system in foggy environment,an improved iterative white balance algorithm is proposed based on some white balance algorithms.Histogram stretching transform is used to extend the dynamic range of the image,the image is transformed from space RGB to space YCbCr,and the white reference points are screened according to the brightness value feature for color correction,the iterative processing of white balance is completed by calculating the cosine similarity of the histogram of the input image and the output image histogram as the termination condition.Experimental results show that the algorithm can effectively improve color bias and color visual effect in foggy images.

关 键 词：雾天图像 亮度特征 直方图拉伸 余弦相似度 迭代白平衡 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]