相位萃取技术可最小化多基线InSAR处理中干涉像对的去相关噪声,已成为雷达干涉测量研究领域中热点技术之一。本文首先设计并实现了一套专门针对小数据集的基于相位萃取技术的多基线InSAR处理流程。整个流程分为两大模块。第一个模块为称为预处理模块,通过图像配准、地形相位移除等操作生成相位萃取技术所需的零基线SAR图像堆栈。第二个模块首先基于多视像元生成相干矩阵,并且在此基础上完成相位萃取操作。然后,利用小基线集技术的思想在萃取相位上分离出各相位稳定点上的形变情况。为了保证结果可靠性,在这一模块中还加入了大气相位预估与移除的步骤。最后,该流程被应用到了5景拍摄于太原地区的PALSAR图像上。实验结果表明,相位萃取技术在小数据集情况下仍然能够有效提升数据的相干性,这有利于多基线InSAR输出结果密度的提升。 Phase extraction technique can minimize the decorrelation noise of interferometric image pairs in multi-baseline InSAR processing and has become one of the hotspot technologies in the field of radar interferometry. This paper first designs and implements a set of multi-baseline InSAR processing flow based on phase extraction technology, which is aimed at small data sets. The whole process is divided into two modules. The first module, known as a preprocessing module, generates the zero-baseline SAR image stack required for phase extraction techniques through image registration, terrain phase removal and other operations. The second module first generates the coherence matrix based on the multi-view elements, and completes the phase extraction operation based on it. Then, using the idea of ​​a small baseline set technique, the deformations at each phase stable point are separated in the extraction phase. In order to ensure the reliability of the results, steps of estimating and removing the atmospheric phase are added to this module. Finally, the process was applied to a five-scene PALSAR image taken in Taiyuan. The experimental results show that the phase extraction technique can still effectively improve the data coherence in the case of small data sets, which is conducive to enhance the density of multi-baseline InSAR output results.