一种面向时间序列InSAR的不连通子网快速连接方法

摘要

时间序列 InSAR 技术目前已广泛应用于地表形变监测，尤其是稳定点目标较多地区的城市地面沉降监测.然而，实际应用中受大气相关距离限制以及模型相干系数阈值影响，连接点目标的Delaunay三角网常常不能完整地连接所有点目标，出现若干个不连通的子网.这种现象导致无法获取部分点目标的形变信息，特别是在非城镇地区影响尤为明显.为了提取大区域地表形变信息，本文提出了一种多层级、不同步长的子网最近邻点目标快速连接方法.其显著特点是通过逐层联网的方式，可快速减少子网数目.与现有复杂网络连接方法相比，该方法在保证结果精度的前提下，数据处理耗时仅为前者的32．56％.%Nowadays,times series interferometric synthetic aperture radar (InSAR)technique has been widely used in ground deformation monitoring,especially in urban areas where lots of stable point targets can be detected.However,in standard time series InSAR technique,affected by atmospheric correlation distance and the threshold of linear model coherence,the Delaunay triangulation for connecting point targets can be easily separated into many discontinuous subnetworks.Thus it is difficult to retrieve ground deformation in non-urban areas.In order to monitor ground deformation in large areas efficiently,a novel multi-layer subnetwork connection (MLSC)method is proposed for connecting all subnetworks.The advantage of the method is that it can quickly reduce the number of subnetworks with valid edges layer-by-layer.This method is compared with the existing complex network connecting mehod.The experimental results demonstrate that the data processing time of the proposed method is only 32.56% of the latter one.