噪声干扰下的雾天降质图像增强方法  

A novel enhancement method for the fog-degraded images with noise computer engineering and applications

作　　者：高明[1,2,3] 邹耀斌[1,2,3] 雷帮军[1,2,3] 徐光柱[1,2,3] 

机构地区：[1]三峡地区地质灾害与生态环境湖北省协同创新中心,湖北宜昌443002 [2]湖北省水电工程智能视觉监测重点实验室,湖北宜昌443002 [3]三峡大学计算机与信息学院,湖北宜昌443002

出　　处：《微型机与应用》2016年第6期39-43,47,共6页Microcomputer & Its Applications

基　　金：国家自然科学基金(联合基金)重点项目(U1401252);国家自然科学基金项目(61272237;61402259;41172298;61202141);湖北省自然科学基金面上项目(2014CFB681);三峡大学楚天学者基金(KJ2012B001);水电工程智能视觉监测湖北省重点实验室开放基金项目(2014KLA02);宜昌市科学技术研究与开发项目(A13-302a-03)

摘　　要：目前国内外的研究主要围绕雾气这一退化因素提出不同的增强方法。然而,在工程实践中,光电雷达监控系统昼光传感器因同时受雾气和噪声的干扰,导致捕获的图像不但整体偏暗、对比度偏低、色彩失真,而且隐含有噪声。在分析光电雷达监控系统昼光传感器捕获的图像的特性基础上,提出了一种噪声干扰下的雾天图像增强方法。首先利用引导滤波器对原图像进行滤波操作,然后对滤波后的图像进行直方图拉伸变换。在大量实际图像上的实验结果表明,该方法能够有效抑制噪声干扰,同时显著增强降质图像。At present,the studies at home and abroad mainly focus on the fog,a primary factor decreasing image quality,to propose different enhancement methods. In engineering practice,however,the daylight sensors in photoelectric radar surveillance system are easy to be simultaneously affected by fog and noise,which results in the captured images with lower brightness,lower contrast,color distortion and hidden noise.Based on the characteristic analysis of the images captured by daylight sensor,this paper proposed a novel enhancement method for the foggy images with noise. The proposed method first filters the original image with the guided image filter,and then implements a histogram stretching transformation on the filtered image. Experimental results on a large number of real images show that the proposed method can effectively suppress noise while significantly enhances the degraded image.

关 键 词：去雾增强 噪声 引导滤波 直方图拉伸 

分 类 号：P315.69[天文地球—地震学]