基于散射模型和非局部滤波的极化SAR图像质量增强算法

发布时间：2018-12-08 07:12

【摘要】：极化合成孔径雷达(SAR)技术是一种先进的遥感技术,在现代国民经济,尤其军事领域的作用是越来越大。极化SAR技术不仅继承了合成孔径雷达技术的全天时,全天候、高分辨率的优势,并且相比较合成孔径雷达技术极化SAR能够获得多种回波信号,从而可以提供更加全面的散射信息。然而,极化SAR图像中的斑点噪声使极化SAR图像信息的质量大大降低,并且现有的极化SAR图像质量增强技术在滤除噪声的同时不能很好地保持极化SAR图像的散射特性,使其在极化SAR图像的分类,分割,目标检测等的应用中也受到影响。因此,极化SAR图像的质量增强技术在极化SAR图像相关处理研究中比较重要,同时,在长期的研究过程中发现,掌握极化SAR图像像素点间散射特性相似性是比较重要的,对极化SAR图像的质量增强,分类,目标检测等研究意义重大,然而,现有的散射特性相似性度量方法不能很好地度量极化SAR数据的散射特性。本文通过对合成以及真实极化SAR数据进行统计分析,提出一种新的极化特性相似性度量方法,并将其应用到极化SAR图像的质量增强技术的研究中,本文主要工作如下:(1)通过对大量合成的和真实的极化SAR数据的极化特性进行研究,提出了一种新的极化SAR相似性度量方法。该相似性度量方法首先对极化SAR数据进行基于散射模型的极化目标分解得到极化特征向量,将极化特征向量建立三维极化特征空间,同时利用空间特征向量间的夹角以及它们模值的关系计算两个像素点的极化相似性,通过与阈值的比较来决定两个像素点是否相似。通过其验证实验发现,该相似性度量方法能够准确并简便地度量出两个像素点的极化相似性。(2)基于极化相似性的极化SAR图像质量增强算法。我们将极化相似性度量方法应用到极化SAR图像的质量增强算法的研究中,提出了两种基于极化相似性度量方法的极化SAR图像质量增强算法。第一种是基于散射特性的局部滤波算法,该算法首先对每一个待处理像素点进行数据类型分类,判断是否是低散射能量点,是否为散射特性孤立点,并根据其数据类型找到其相似像素点集合,进一步利用Lee滤波模型进行相干斑抑制。第二种是基于散射特性和NL-Lee滤波的极化SAR图像质量增强算法,该算法同样首先对每一个待处理像素点进行数据类型分类,判断像素点是否为散射特性孤立点,并根据其数据类型找到其相似像素点集合,进一步利用NL-Lee滤波算法计算出参数k(i,j)以及利用数据类型来得到权值w(i,j),最终得到极化SAR图像质量增强的结果。
[Abstract]:Polarimetric synthetic Aperture Radar (SAR) is an advanced remote sensing technology, which plays a more and more important role in modern national economy, especially in military field. Polarization SAR technology not only inherits the advantage of all-day, all-weather and high resolution of synthetic Aperture Radar (SAR) technology, but also can obtain many kinds of echo signals compared with SAR technology. Thus more comprehensive scattering information can be provided. However, the speckle noise in polarimetric SAR images greatly reduces the quality of polarimetric SAR images, and the existing enhancement techniques of polarimetric SAR images can not preserve the scattering characteristics of polarimetric SAR images while filtering noise. It is also affected in the application of polarimetric SAR image classification, segmentation and target detection. Therefore, the quality enhancement of polarimetric SAR images is very important in the research of correlation processing of polarimetric SAR images. At the same time, it is found that it is important to grasp the similarity of scattering characteristics between pixels in polarimetric SAR images during a long period of research. It is of great significance to study the quality enhancement, classification and target detection of polarized SAR images. However, the existing methods for measuring the similarity of scattering characteristics can not well measure the scattering characteristics of polarimetric SAR data. Based on the statistical analysis of synthetic and real polarimetric SAR data, a new similarity measurement method of polarization characteristics is proposed in this paper, and applied to the study of quality enhancement of polarimetric SAR images. The main work of this paper is as follows: (1) by studying the polarization characteristics of a large number of synthetic and real polarimetric SAR data, a new polarimetric SAR similarity measurement method is proposed. Firstly, polarization eigenvector is obtained by polarization target decomposition based on scattering model for polarimetric SAR data, and the polarization feature space is established by polarization eigenvector. At the same time, the polarization similarity of the two pixels is calculated by using the angle between the spatial eigenvectors and the relationship between their modes, and the similarity between the two pixels is determined by comparing with the threshold. The experimental results show that the proposed method can accurately and easily measure the polarization similarity of two pixels. (2) the polarimetric SAR image quality enhancement algorithm based on polarization similarity. We apply the polarimetric similarity measure method to the study of the quality enhancement algorithm of polarimetric SAR images and propose two kinds of polarimetric SAR image quality enhancement algorithms based on the polarimetric similarity measurement method. The first is a local filtering algorithm based on scattering characteristics. Firstly, the algorithm classifies the data types of each pixel to determine whether it is a low scattering energy point or a scattering characteristic outlier. The similar pixel sets are found according to their data types, and speckle suppression is further carried out by using Lee filter model. The second is a polarimetric SAR image quality enhancement algorithm based on scattering characteristics and NL-Lee filtering. The algorithm also classifies the data types of each pixel to determine whether the pixel is a scattering characteristic outlier. The similar pixel set is found according to its data type, and then the parameter k (ij) is calculated by using NL-Lee filter algorithm and the weight value w (ij) is obtained by using data type. Finally, the quality enhancement results of polarimetric SAR images are obtained.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52

【共引文献】