非局部均值滤波,SAR图像去噪.docxVIP

基于迭代型非局部均值滤波的SAR图像去斑SAR Image Despeckling Based on Iterative Non-local Means Filter2014年6月15日摘要合成孔径雷达干涉测量（Interferometric Synthetic Aperture Radar, InSAR）技术因具有全天时、全天候、高精度、大区域等突出优点已成为目前最具潜力的对地观测技术之一，在研究地震形变、火山运动、冰川漂移、城市沉降以及山体滑坡等方面表现出极好的应用前景。SAR影像（单视复数影像）既记录了地面目标的辐射强度信息，也记录了与斜距（雷达天线到地面分辨单元）有关的相位信息，InSAR正是利用这些具有高敏感特性的相位信号来分离和提取出有用信息。InSAR特殊的成像机制导致图像相干斑噪声的产生，使得图像信噪比下降，为特征提取造成困难。因此，对SAR图像去除相干斑噪声是InSAR数据处理中不可缺少的步骤。本文深入研究了SAR图像噪声的产生机理和模型，分析了几种较为经典的空域和频域滤波算法，指出这些传统算法所存在的问题：滤除噪声的同时模糊了图像的边缘纹理信息。针对传统滤波的缺点，基于非局部均值算法思想，提出了迭代型非局部均值滤波算法，充分考虑了SAR图像的噪声模型，修改了原有的权值函数，通过多次迭代来增加相似性的准确度。最后，采用本文算法依次对合成SAR图像和真实SAR图像进行滤波，实验表明，即使是纹理复杂、边缘信息丰富的图像，迭代型非局部均值滤波也能得到非常好的效果。将该算法同Boxcar滤波、精制Lee滤波、IDAN滤波进行比较，噪声抑制方面效果相当，但细节保持方面本文算法更加突出。关键字：SAR图像去斑非局部均值滤波斑点噪声迭代AbstractCurrently, Interferometric Synthetic Aperture Radar (InSAR) has become one of the most potential techniques for earth-observing, owing to its prominent merits, such as all-time, all-weather, high accuracy and large observation areas, etc. This technique indicates the great prospect in aspects of earthquake deformation, volcanic movement, glacial drift, urban settlement and landslide.SAR images (Single-Look Complex, SLC) record the radiation intensity information of ground target, as well as phase information related to slant-range (distance between radar antenna and ground resolution cell), InSAR exactly use the phase with a high sensitive feature to separate and extract useful information.The special imaging mechanism of SAR leads speckle noise, which cause decreasing of signal to noise ratio (SNR) and difficulties in research of target recognition and feature extraction. Therefore, despeckling is an indispensable process during InSAR data processing.This paper deeply studies the mechanism and model of SAR image noise, analyses of several classic spacial and frequency domain filtering algorithm, Points out the existing problems in the traditional algorithm, poor detail preserving.A new method, iterative Non-local Means filter, was proposed