基于相干曲面移动拟合的SAR影像高精度配准方法

SAR image registration with high accuracy based on moving coherence surface fitting

本文采用斜距-多普勒模型与卫星轨道状态矢量探查影像粗配准点,以粗配准像素中心及其邻域相干系数阵列拟合相干曲面,通过快速求解二元连续相干函数的极值点作为影像精细配准位置,建立主从影像同名像素的坐标映射转换模型。实验分别以我国东部平原地区的ERS SAR影像和西部龙门山地区的ALOS PALSAR影像为例,进行InSAR影像配准与干涉计算,结果表明,SAR影像配准后的相干性提高约5%。

In the paper, the satellite orbital state vector and slang-range Doppler model were first used to seek the coarse matched pixel. The mathematical model was fitted based on the coherence values of central pixel and its adjacent pixels. The accurate position of matched point was obtained by computing the maximum value of fitted surface function. The matched homologous points were em- pXoyed to establish the coordinate transformation relationship between master acquisition and slave image. The ERS-1/2 SAR images covering Shanghai plain area and ALOS PALSAR images covering Longmenshan mountainous area were respectively used in the study. The testing results showed that the coherence could be improved by about 5%.