基于暗通道先验和四叉树空间索引的图像去雾算法

Image Defogging Algorithm Based on Dark Channel Prior and Quadtree Spatial Index

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.