摘要随着计算机视觉的快速发展，我们对户外成像系统所获得图片的质量的要求也越来越高。薄云薄雾天气下，图像会发生严重的退化现象，对比度降低，动态范围减小，颜色偏灰白，整体辨识度降低。78049

本文结合大气成像模型对薄云薄雾天气图像退化的原理进行了分析,说明了雾天图像降质的原因，并利用基于图像处理的非模型增强手段对雾天图像进行了清晰化处理。首先采用全局直方图均衡化算法处理图像并对其进行加修正处理，获得了比较好的去雾效果，但同时也会造成图像的部分细节缺失和对比度过度增强。继而，本文着重采用了Retinex算法实现去薄云薄雾，包括基于中心/环绕的SSR、MSR、MSRCR和基于像素比较的Frankle_McCann、McCann99，实现了比较好的色调重现和动态压缩范围增强，并对采用不同算法处理后的图像的质量进行参数运算与对比分析，得出各算法的优缺点。

毕业论文关键词 薄云薄雾图像增强 直方图均衡 Retinex算法

Title   Research on image processing under weather condition of thin cloud or mist 

Abstract With the rapid development of computer vision, our requirements for the quality of the image obtained by the outdoor imaging system are more stringent than before。 In the weather with thin cloud or mist,the images will deteriorate。Their contrast will be reduced, and their dynamic range will decrease。What’s more,the picture will turn gray and the overall recognition will become lower。

First,we analyse the mechanism of degradation of images under weather condition of thin cloud or mist combining with the atmospheric imaging model。In this way,we illustrate the reason for decline of images’quality and use non-model algorithms based on image enhancement to make the images clear。 The global histogram equalization is used in this thesis to process image with modificatory factor and it performs well, but it may cause loss of the image details and excessive contrast enhancement。Then we focus on the Retinex algorithm which includes SSR、MSR、MSRCR、Frankle_McCann and McCann99, and we achieve better color rendition and dynamic range compression。 Finally,we calculate and compare the image quality parameters of different algorithms to analyse their own strengths and weaknesses。

Keywords thin cloud and mist image enhancement histogram equalization Retinex algorithms

目录

1 绪论 1

1。1引言 1

1。2课题研究的背景及意义 1

1。3国内外研究现状 2

1。4本文章节内容安排 3

2 薄云薄雾天气图像退化机理 4

2。1薄云薄雾天气的形成及影响 4

2。2大气散射理论 5

2。2。1入射光衰减模型 5

2。2。2大气光成像模型 6

2。3雾天图像退化原因与特性 7

2。4去雾图像的评价标准 7

3 基于直方图均衡化的去薄云薄雾算法 9

3。1直方图均衡化基本原理