薄尘雾退化图像的处理研究

【作者】 李鹏；

【导师】 孙锦涛；

【作者基本信息】 南京理工大学 ， 通信与信息系统， 2005， 硕士

【摘要】 薄尘雾等恶劣天气条件下获得的图像受到了严重的退化,这大大降低了图像的应用价值,对军事、交通、遥感等各领域的图像信息处理造成了极大了影响,研究如何对薄尘雾退化图像进行有效地处理有着非常重要的现实意义。 本文首先介绍了尘雾等恶劣天气对图像的退化效应,对Oakley退化模型和基于大气传递函数的退化模型进行了分析,并对基于退化模型的复原方法作了讨论,指出其在实际应用中遇到的困难。进而在对薄尘雾退化图像特征分析的基础上,从图像增强的角度探讨简单而有效的处理方法。 对于薄尘雾退化的彩色图像,用直方图均衡和同态滤波等常规增强算法处理往往不能得到满意的效果。如果将薄尘雾退化图像看作由尘雾等大气因素综合作用的部分和反映场景本来特征的部分组成,对这类图像的处理就是提取反映场景本来特征的部分,这为应用Retinex理论进行处理提供了可能。接下来重点对将Retinex算法应用于薄尘雾退化图像的处理进行了探讨。简要介绍了Retinex理论的基本思想和算法发展,从实际应用出发,选择了其中三类常用而易于实现的算法:基于像素比较的Retinex算法,SSR和MSR(CR)算法,Kimmel Retinex算法。在分别将不同类型的Retinex算法应用于处理薄尘雾退化图像时,对算法进行了相应的改进,取得了满意的处理效果。更多还原

【Abstract】 The image we obtained in bad weather conditions is often significantly degraded by fog and mist. It greatly decreased the use of the image and make bad effect on the information processing in military, traffic, remote-sensing etc. As a result, the research on how to effectively process this kind of degraded images has very important meaning.First the degradation effect on the image by bad weather is presented, and the Oakley degradation model and the degradation model based on the atmosphere transfer function are analyzed. The restoration methods based on the degradation model are discussed in this paper, and its difficulties we met in practical application are also indicated. Based on the characteristic analysis of the image degraded by fog and mist, the simple and effective processing methods from the view of image enhancement are discussed.Satisfied results usually can not be got when processing color images degraded by fog and mist with the conventional enhancement algorithms such as histogram equalization and homomorphic filtering. If the image degraded by fog and mist is thought to be composed of two parts: One is the part effected by the atmosphere, and the other is the essential character of the scene, the process is to obtain the second part. This offers the probability to use Retinex theory. Then the paper lays strong emphasis on the discussion of the application of Retinex algorithm for processing the images degraded by fog and mist. The basic idea and the development of the Retinex algorithm are briefly introduced. For practical application, three kinds of algorithms which are Retinex algorithm based on pixel comparing , SSR and MSR(CR), Kimmel Retinex are chosen for their common use and easy implementation. These algorithms are improved when used for processing the images degraded by fog and mist. Satisfied results are got.更多还原