一种InSAR相关系数精估计方法

摘要

The InSAR correlation,a measure of the similarity of interferometric SAR image pairs, provides an elemental parameter for InSAR applications. In practice,correlation is observed by comparing the radar return across several nearby radar images pixels. The estimation is biased by underlying topographic interferometer fringe and finite pixel samples. We present here a method for cor recting bias in InSAR estimation,which is resulting in significantly more accurate correlation. Firstly,we correct the underlying topographic fringe patterns by phases simulated by existing rough DFMs. A second kind statistics method is then applied to bias removal. We demonstrate the value of our method using data collected over the Etna volcano by the SIR-C/XSAR sensors.%InSAR的相关系数是对SAR干涉图像对相似度的度量,是InSAR技术应用分析的一项基本参数.相关系数的估计是通过比较两幅SAR图像中对应位置的相邻像素值得到的,由于采样点数目的限制及存在由于地形起伏引起的干涉相位,估计结果往往与真实值相差较大.本文提出了一种能够大幅减小相关系数估计误差的方法,首先利用粗精度DEM仿真地形相位以消除地形起伏对相关系数估计的影响,随后采用第二类统计方法进一步提高了相关系数的估计精度.我们用SIR-C/XSAR的Etna火山干涉数据对本文方法进行了验证.