摘要当今社会图像采集在各个领域的重要性越来越显得突出，例如在交通管制方面对违规车辆的拍照，以前室外各种重要位置的监控等方面，但是随着霾，雾现象的出现，图像的采集会变得困难。通常就会引起所得图像的对比度会有所降低，颜色也会出现失真，或者模糊等质量的退化问题。这会使得对目标特征的判定和对目标的跟踪等活动没有办法正常的进行下去。所以对于如何给雾天图像去雾是具有实际研究价值的。具体的工作如下：73354

  先对雾天图像的降质原因、过程以及特点都进行细致的分析，主要研究直方图均衡化以及多尺度Rrtinex增强算法来给图像去雾，并由此发现这一方法在场景灯光和环境光比较相近的区域会出现自身的不足，从而使得去雾结果出现问题，主要体现在局部对比度比较低以及细节不显著。为了提高图像的质量，采用改进后的图像去雾方法，二阶段图像去雾方法，使用直方图均衡化在暗原色先验之后再对图像进行局部的对比度修正。实验的结果可以看出，这种方法可以有效克服使用暗原色先验带来的缺陷，很好的提高雾天采集到图像的对比度以及清晰度。

毕业论文关键词   双边滤波  大气散射模型  直方图均衡化  多尺度Retinex算法 

毕业设计说明书外文摘要

Title    Study on the Haze Removal Algorithm and Implementation                     

Abstract In today's society the stressing the importance of image acquisition in all fields, such as pictures of illegal vehicles in traffic control, important position before all kinds of outdoor monitoring, etc。, but as the haze and mist phenomenon is the emergence of the image acquisition becomes difficult。 Often can lead to lower contrast of the image and color distortion, and the fuzzy quality degradation。 It will lead to determine the features of the target and the target tracking activities such as there is no way to normal。 So for this type of image to fog research has practical application value。

This paper is to study the method of single image to fog, through experiments to verify this to the effectiveness of the fog。 Specific work is as follows:

First mass reduction for image contrast, cause, process and characteristics are analyzed, the main research dark grey prior image to fog method, and found that this method in the light of the scene can appear close to the ambient light in the area is small, so as to make the fog to result the problem, is embodied in local contrast is not high, and details is not obvious。 In order to improve the quality of the image, using the improved image to fog method, two-phase image to fog method, using the histogram equalization after dark grey prior to local contrast of image correction。 Results it can be seen that this method can effectively overcome the defects resulted from using the dark color prior, very good to the image contrast and definition of fog collection。

Keywords  Bilateral Filtering  Atmospheric Scattering Model  Histogram Equalization  Multiscale Retinex Algorithm

目录

1。引言 1

1。1研究背景及意义 1

1。2国内外研究现状 1

1。2。3存在的问题以及研究的难点 2

1。3本文的主要工作 2

2图像退化机理 3

2。1雾的形成 3

2。2大气散射模型 3

2。2。1直接衰减模型