顾及时空特性的 SBAS 高质量点选取算法

摘要

Traditional coherence-based point-selection method in the SBAS technique is often suffered from the side lobe effect problem,which can result in the omission of partial high-qual ity points and reservation of partial low-qual ity points.A new algorithm to select high-qual ity points is proposed in the SBAS technique by quantifying the separated noise phase component.Compared with the traditional fi ltering methods,it is found that the Non-Local algorithm considering the homogeneous points can accurately separate spatial ly-correlated phase from the unwrapping differential interferometric phase,and then the precision of the noise phase component estimation is improved.Coherence-based approach and our new algorithm are compared by selecting high-qual ity points and mapping the surface deformation of the test area in Shanghai.There it is uti l ized the 24 JERS-1 SAR images acqui red from 1992 to 1998.The results suggest that the proposed algorithm not only can select the high-qual ity points in fields covered by farmlands and hamlets which are omitted by coherence-based method,but also can effectively exclude the points extracted by coherence-based method due to the side lobe effect.%SBAS 技术中常用的相干系数选点方法受旁瓣效应的影响，容易造成部分高质量点的漏选以及低质量点的误选。针对这一问题，本文提出了顾及时空特性的 SBAS 高质量点选取算法，根据干涉相位各个成分的空间特性，分离出噪声相位判别点的质量。试验表明考虑同质点的 Non-Local 滤波能够更加可靠地提取干涉相位中的空间相关相位，从而提高噪声相位估计的准确性。利用覆盖上海地区1992—1998年的24景 J ERS-1影像作为试验数据，分别用相干系数法以及本文的方法进行选点。结果表明本文算法能够有效地选出相干系数法在农田与村落交错地区漏选的高质量点，同时排除了相干系数法由于旁瓣效应影响的误选点。