一种简单快速的图像去雾算法  

A SIMPLE AND FAST IMAGE DEHAZING ALGORITHM

作　　者：刘巧玲[1] 张红英[1] 林茂松[1] 

机构地区：[1]西南科技大学信息工程学院,四川绵阳621010

出　　处：《计算机应用与软件》2013年第7期71-74,共4页Computer Applications and Software

基　　金：国家自然科学基金青年基金项目(60802040);四川省科技厅基金项目(2011JQ0041;11zs2009)

摘　　要：在雾、霾等天气条件下,大气粒子的散射作用导致捕获的图像严重降质。提出一种简单快速的基于物理模型的图像去雾新算法。首先从雾形成的角度出发,考虑到传播图像的低频性和单通道性,对大气散射模型进行变型化简,得到新的去雾模型。然后,利用暗原色先验方法估计大气光值A,并带入新的简化模型,得到去雾图像。方法不需要求取过多的变量,只需要求取大气光值A,大大提高了算法的效率。最后大量实验表明,该算法无论在处理速度还是去雾效果上都优于现有算法。The scattering function of atmospheric particles in bad weather conditions such as haze, fog and mist will significantly degrade the images captured. In this paper, we propose a simple and fast physical model-based algorithm for single image dehazing. Firstly, taking into account the low-frequency and single channel of image transmission, we simplify the atmospheric scattering model from the perspective of fog formation and obtain new dehazing model. Then, we use dark channel prior method to estimate the atmosphere light value A, and take into the new simplified model to get the dehazing image. This method does not need to deal with too much variables but only calculates the atmosphere light value A, this greatly improves the efficiency of the algorithm. Finally a large number of experiments show that the proposed algorithm has better performance in image dehazing quality and computational time than the current algorithms.

关 键 词：图像去雾 物理模型 暗原色先验 大气光值A 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]