引导滤波在雾天图像清晰化中的应用  

Application of Guided Filter to Haze Image Sharpening

作　　者：王伟鹏[1] 戴声奎[1] 

机构地区：[1]华侨大学信息科学与工程学院,福建厦门361021

出　　处：《华侨大学学报（自然科学版）》2015年第3期263-268,共6页Journal of Huaqiao University(Natural Science)

基　　金：福建省科技计划重点基金资助项目(2013H0030);中央高校基本科研业务费;国家自然科学基金培育计划专项基金资助项目(JB-ZR1145)

摘　　要：根据大气散射物理模型,分析并解决单幅图像去雾的关键问题,以消除雾气对室外机器视觉系统的影响.实验结果表明:利用引导滤波器的局部平滑特性估计大气光幕,可有效克服光晕效应和颜色过饱和现象;通过计算雾气最浓区域的平均值获取大气光强度值,可解决部分图像偏色问题;对复原的图像进行自适应增强处理,提高了图像的整体视觉效果.通过与多种典型的图像去雾算法的比较,说明该方法能更有效地消除图像中的雾气、真实复原场景的对比度和颜色,同时具有很快的执行速度,有利于算法的实时实现.According to the physical model of atmospheric scattering,this paper analyses and addresses major problems for single image haze removal to eliminate the effects of fog or haze for outdoor machine vision system.The experimental results show that the proposed method overcomes the limits of halo artifacts and over-saturation through estimating the atmospheric veil using the local smoothing characteristics of guided filter.By computing the average values of the most dense haze to obtain the global atmospheric light intensity,this algorithm could adjust the color cast of the image.Final-ly,a post-processing procedure is adopted to enhance the visual effect of the recovered image.As compared with typical haze removal algorithms,the proposed method achieves good restoration for contrast and color fidelity more effectively and allows a very fast implementation to achieve real-time.

关 键 词：雾天图像 引导滤波 去雾算法 大气散射模型 大气光幕 图像增强 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]