摘要
本文针对暗通道先验算法中大气光值估计不准确的缺点,采用了一种用四叉树层次分解估计大气光,然后进行图像去雾的方法。首先,用四叉树分割对原图像进行分割,得到4个区域;其次,选定像素均值减去所对应的方差结果最大的区域进行递归操作;最后,计算出大气光值.实验结果表明,本文算法能在保证去雾效果的前提下,运用四叉树大气光值估计得到改进后的大气光值,从而弥补暗通道先验算法当中的不足之处.
Aiming at the disadvantage of inaccurate estimation of atmospheric light value in dark channel prior algorithm,this paper adopts a method of estimating atmospheric light with quadtree hierarchical decomposition,and then carries out image defogging.Firstly,the original image is segmented by quadtree to get four regions;Secondly,recursive operation is performed in the region where the result of subtracting the mean value from the corresponding variance is the largest;Finally,the atmospheric light value is calculated.The experimental results show that this algorithm can ensure good defogging effect,and the improved atmospheric light value can be obtained by using quadtree atmospheric light value estimation,so as to make up for the shortcomings of the dark channel prior algorithm.
作者
周昊
廖洪波
李婷
姜龙辉
高铃
ZHOU Hao;LIAO Hongbo;LI Ting;JIANG Longhui;GAO Ling(School of Big Data and Artificial Intelligence,Chizhou University,Chizhou Anhui 247000,China)
出处
《信息与电脑》
2021年第10期57-59,共3页
Information & Computer
基金
安徽高校自然科学研究重点项目(项目编号:KJ2019A0864)
安徽省大学生创新创业训练计划项目(项目编号:202011306075)。
关键词
图像去雾
暗通道先验
四叉树多层次分解
image defogging
dark channel prior
quadtree multi-level decomposition