结合场景特征的雾天道路图像清晰化  

Road Image Sharpening in Foggy Condition Based on Scene Features

作　　者：王伟鹏[1] 项文杰[1] 戴声奎[2] WANG Weipeng;XIANG Wenjie;DAI Shengkui(Minnan Science and Technology Institute, Fujian Normal University, Quanzhou Fujian 362332, China;College of Information Science and Engineering, Huaqiao University, Xiamen Fujian 361021, China)

机构地区：[1]福建师范大学闽南科技学院,福建泉州362332 [2]华侨大学信息科学与工程学院,福建厦门361021

出　　处：《莆田学院学报》2017年第2期28-32,共5页Journal of putian University

基　　金：福建省科技计划重点项目(2013H0030);中央高校基本科研专项(JB-ZR1145)

摘　　要：为解决大雾天气下道路图像的降质模糊问题,提出一种基于成像模型的清晰化算法。由于道路图像的景深有别于一般的自然场景,分析并总结道路场景基本特征可用于修正成像模型中部分参数的估计方式:一是利用简化的高斯平滑滤波器获取大气耗散函数;二是引入局部还原控制因子,实现不同景深的自适应清晰化增强;三是准确定位天空区域,选取像素平均值作为天空亮度。实验证明,结合场景特征的清晰化算法能够有效去除雾气的不良影响,复原结果的整体亮度得到有效保持,视觉效果真实自然。In order to deal with the problem of road image degraded and blurred in foggy condition,a sharpeningalgorithm based on an imaging model is proposed.Since the scene depth of road image is different from generalnatural scenes,some essential features are analyzed and summarized to improve parametric estimation method fromthe imaging model:firstly,the atmospheric veil is obtained by a simplified Gaussian filter;secondly,the controlfactor about local restoration is introduced to achieve adaptive sharpening and enhancement according to differentscene depth;thirdly,the sky areas are accurately positioned,therefore the global atmospheric light can be estimatedby calculating an average.Experimental results show that the proposed sharpening algorithm based on scene featuresremoves fog and keeps the global brightness of recovered result more effectively,and then achieves better visualeffect.

关 键 词：道路图像 清晰化 去雾 图像增强 参数自适应 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]