一种基于多尺度Retinex算法的图像去雾方法  

AN IMAGE DEFOGGING METHOD BASED ON MULTI-SCALE RETINEX

作　　者：罗会兰[1] 林家彪[1] 

机构地区：[1]江西理工大学信息工程学院,江西赣州341000

出　　处：《计算机应用与软件》2013年第4期58-60,127,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(61105042);江西省自然科学基金项目(2010gzs0075);江西省教育厅科技项目(GJJ11464;GJJ11465)

摘　　要：在有雾天气条件下拍摄的图像,由于大气的散射作用,导致图像的内容模糊不清,对比度下降,给交通系统及户外视觉系统的应用带来严重的影响。通过改进大气物理退化模型,在多尺度Retinex算法MSR(Multi-Scale Retinex)的基础上,提出一种新的去雾方法。该方法根据雾天成像机理对图像建立模型,再根据MSR算法对建模后的图像进行处理。实验表明,该方法能有效去除雾化效果,实现彩色退化图像的复原。Images taken in foggy weather are blurred in view and with poor contrast because of atmosphere scattering,these seriously impact the application of traffic system and outdoor visions system.By improving the degradation model of atmospheric physics and based on multi-scale Retinex（MSR） algorithm,we propose a new defogging method.This method sets up the model for images according to the imaging mechanism in foggy weather,and then processes the images being modelled based on MSR algorithm.Experiments demonstrate that the method can effectively eliminate the foggy impact on the image and realises the recovery of colour degraded images.

关 键 词：去雾 大气物理模型 多尺度RETINEX算法 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]