摘要
针对雾霾恶劣天气状况下获取的图像视觉效果差,提出了一种基于视觉感知的快速雾天图像清晰度复原方法,以测算大气光学物理模型的两个重要参量。首先采用阈值分割结合二叉树分割的方法拟合较为精准的大气光值,进而采用自适应各向异性型高斯滤波与色调调整方法优化透射率。与已有方法对比实验结果表明,所提算法的去雾效果图饱和清晰,能够保留清晰的边缘细节和较高的对比度,算法的处理效率高,能满足实际应用需求。
In view of the poor visual effect of images obtained bad weather conditions of heavy haze,this paper proposed a fast restoration method of haze images based on visual perception,and measured two important parameters of atmospheric optical physical model.Firstly,this paper used threshold segmentation combined with binary tree segmentation method to fit a relatively accurate atmospheric light value,and then adopted adaptive Gaussian filtering and tone adjustment methods to optimize the transmittance.The experimental results show that in comparison with the existed algorithms,the effectiveness of the deha- zing results of the proposed algorithm is saturated and clear.Besides,it also can retain clear edge details,high contrast,high processing efficiency and can meet the actual application requirements.
作者
付辉
吴斌
张红英
Fu Hui;Wu Bin;Zhang Hongying(School of Information Engineering,Southwest University of Science & Technology,Mianyang Sichuan 621010,China)
出处
《计算机应用研究》
CSCD
北大核心
2019年第8期2522-2526,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(61401072)
四川省教育厅重点项目(15ZA0118)
关键词
视觉感知
阈值分割
二叉树
高斯滤波
色调调整
visual perception
threshold segmentation
binary tree
Gaussianfiltering
tone adjustment
