Identification and estimation of InSAR interferogram noise based on wavelet transform

CAI Guolin; LIU Guoxiang; ZHANG Aoli; SUN Meiling

doi:10.11834/jrs.20143120

Views : 0 下载量: 756 CSCD: 1

PDF
Export
Share
Collection
Album

Identification and estimation of InSAR interferogram noise based on wavelet transform
Vol. 18, Issue 3, Pages: 537-546(2014)
Published：2014
DOI： 10.11834/jrs.20143120
稿件说明：

移动端阅览

[1]蔡国林,刘国祥,张奥丽,孙美玲.基于小波域的InSAR干涉图噪声识别与估计[J].遥感学报,2014,18(03):537-546. DOI： 10.11834/jrs.20143120.

CAI Guolin, LIU Guoxiang, ZHANG Aoli, et al. Identification and estimation of InSAR interferogram noise based on wavelet transform[J]. Journal of Remote Sensing, 2014, 18(3): 537-546. DOI： 10.11834/jrs.20143120.

摘要

为识别InSAR干涉图噪声的类型

从而更有效地进行干涉图处理

提出一种基于小波变换的干涉图噪声类型识别方法。该方法结合小波变换和对数变换算法

从能量和分布特征两个角度

对经小波变换后的InSAR干涉图小波系数的特点进行分析。首先研究低频子带小波系数在对数变换前后的能量分布特点

以此识别干涉图的噪声类型

然后在此基础上利用高频子带系数估计噪声的大小。分析结果显示

干涉图中的噪声为加性噪声

且干涉图高频系数呈高斯正态分布

其噪声大小可用尺度1上的HH子带系数估计。这些结果对在小波域内进行InSAR干涉图噪声的分析与处理具有一定指导意义。

Abstract

To identify the type of noise in an Interferometric Syntheic Aperture Radar（InSAR） interferogram and facilitate more effective data processing

especially for interferogram filtering based on wavelet transform

we developed a novel method for identifying the noise type and estimating the noise level. This method combines wavelet transform with logarithmic transformation and includes an in-depth analysis of the characteristics of interferogram wavelet coefficients from the viewpoints of energy distribution and statistical interpretation. First

we investigated the energy distribution characteristics of low-frequency subband wavelet coefficients before and after the logarithmic transformation to identify the noise type of the interferogram. On this basis

the noise level could be estimated using high-frequency subband wavelet coefficients. From the analysis results of this study

we conclude that the noise type of the InSAR interferogram is additive noise

the high-frequency coefficients show a Gaussian distribution

and the noise level can be estimated by a high-frequency subband on one scale. These results provide further scope for noise analysis and p rocessing of InSAR interferograms based on wavelet transform.