合成孔径雷达(Synthetic Aperture Radar, SAR)是一种高分辨率的成像雷达,具有全天候、全天时的工作能力,在军事应用和国民经济的各个领域中表现出巨大的潜力和应用前景。由于 SAR 图像存在着显著的相干斑噪声,减弱了 SAR 系统对目标的分辨能力,极大地降低了传统的边缘检测、图像分割、目标分类等信息扩展技术的有效性,增加了SAR图像的解译工作的难度,因此必须对相干斑进行抑制。 一个理想的去斑算法应该在平滑的同时保持图像的边缘等细节不受损失,目前存在各种各样的算法,但没有一种方法能够完美的满足这一要求。 本文首先简要介绍了研究相干斑抑制算法的目的和意义、目前的研究状况,以及本文的主要工作。然后讨论了相干斑的形成原理,分析了其统计特性,研究了相干斑的噪声模型。介绍了目前常用的几种相干斑抑制算法,详细分析了算法原理并初步评价了其优缺点,重点对其中的几何滤波算法作了改进。最后,建立了一系列的图像质量评估指标,通过对真实SAR图像进行处理,对上述几种算法的优劣作了定量的分析和比较,最终对这些算法的效果和优缺点作了较为客观的评价。本文还利用 Visual C++编程实现相干斑抑制算法,并将滤波算法模块化,可以在实际应用中方便使用这些滤波算法。8338
关键词 合成孔径雷达;相干斑抑制;空域滤波;马尔可夫随机场
Title Speckle smoothing algorithm for SAR images
Abstract
Synthetic Aperture Radar (SAR) is a high- resolution imaging radar, It can work under
any weather at any time and has shown great potential and prospects in military
applications and allareas of the national economy. As the significant speckle noise in
SAR image which weakening the SAR system resolution capability of the target, and
greatly reducing the traditional edge detection, image segmentation etc, increasing the
difficulty of SAR image interpretation, if is necessary to suppress the speckle.
An ideal algorithm should smooth the speckle without blurring edges and fine
detai l. But most algorithms cannot satisfy these two demands very good.
In the preface to this article, first, the basic purpose and significance of speckle
smoothing algorithm for SAR images are introduced briefly. Secondly, we look back on
the history of the speckle reduction algorithm. And finally, the principal contribution of
this dissertation is summarized here. Then, described how the speckle formed, and
discusses the noise model thus the fundamental properties of speckle in SAR images.
Several well-known speckle filtering algorithms are discussed. The concept of each
filtering algorithm is discussed in detail and we evaluate their performance. Then
improve some of the filtering algorithms. To compare speckle smoothing algorithms
quantitatively, a set of performance criteria is established at last. And tests are
performed on real SAR images. Then comparisons are made for the effectiveness of
these algorithms in speckle reduction and edge, point target gontrast preservation.
In this paper, the algorithm is programmed by Visual C++. The filtering algorithm is
modular and can be easily used in practical applications.
Keywords SAR; speckle noise; spatial filter; Markov Random Field
目 次