基于直方图规定化的图像去雾算法  

Image Haze Removal Algorithm Based on Histogram Specification

作　　者：王万国[1,2] 王滨海[1,2] 张晶晶[1,2] 李丽[1,2] 

机构地区：[1]国网山东省电力公司电力科学研究院国家电网公司电力机器人技术实验室,山东济南250002 [2]山东鲁能智能技术有限公司,山东济南250101

出　　处：《计算机技术与发展》2014年第9期241-244,共4页Computer Technology and Development

基　　金：国家电网科研基金项目(2012A-17)

摘　　要：直方图规定化是图像增强领域一个常用的算法,文中提出一种通过高斯函数加权的直方图规定化的图像去雾算法。首先通过分析晴天与雾霾天气下图像的直方图的特点,提出一种通过对高斯函数中方差的改变和高斯函数的加权的方式,解决了原有的单纯高斯函数直方图规定化图像偏暗的问题。通过实验图像的对比可以看出,文中提出的算法可有效去除雾霾天气的影响,其处理效果明显优于直方图规定化算法,而且计算量小、处理速度快、不需要人工干预。Histogram specification is a commonly used algorithm in image enhancement field. Propose an image haze removal algorithm of histogram specification based on the weighted Gaussian probability density function （Gaussian PDF） in this paper. Firsfly, analyzing the characteristics of image histogram that captured in sunny, fogging and haze weather. Then, solve the weak intensity problem of image specification of the single Ganssian function through changing the variance and weighted Gaussian PDF. The experimental results show the algorithm is able to remove the fog effectively, which is superior to the some existing algorithms of histogram specification about efficiency. It also has many advantages such as low computation ,fast processing speed, no manual intervention.

关 键 词：高斯函数 直方图规定化 图像去雾 加权 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]