城市地表及基础设施形变监测对城市安全运行具有重要意义。合成孔径雷达干涉测量技术（InSAR）是城市形变监测的重要手段之一。在复杂城市场景中，传统永久散射体InSAR技术（PS-InSAR）易受大气延迟影响、高估相位噪声水平；基于PS弧段的InSAR技术（PSP-InSAR）通过解算点对参数，可有效提取高质量监测点，但在算力有限时，点对搜索覆盖不足。为此，本文提出一种基于密集弧段连接的时序InSAR方法（Exhaustive Pairs InSAR, EP-InSAR），适用于存在局部非线性形变的复杂城市场景。该方法采用顺次基线连接策略，抑制时空双差相位形变分量，允许仅开展一维弧段参数解算，显著提高计算效率；进一步设计双阈值驱动的迭代网络扩张算法，实现对高质量点对的近似穷举，减弱对相位质量幅度先验的依赖，充分发掘稳定监测点。最终，剔除高程与热胀冷缩项获得低频干涉图序列，提取形变时间序列产品。上海地区TerraSAR-X数据实验结果表明，EP-InSAR方法有效提升了监测点覆盖率，能够筛选出大量不符合传统幅度先验的高质量监测点，其中29.3%监测点的振幅离差指数大于0.6；在3000*3000的处理窗口内，提取由6000万条高质量弧段连接的91万监测点，为监测点参数解算提供大量多余观测；最终，发现多处可与卫星历史影像印证的局部形变隐患点。整体算法流程已初步实现并行优化，具备工程应用可行性，有望在城市建筑形变普查和健康监测中应用推广。

Monitoring urban ground and infrastructure deformation is critical for the safe operation of cities. Interferometric Synthetic Aperture Radar (InSAR) has become a vital tool for this purpose. In complex urban environments, conventional Persistent Scatterer InSAR (PS-InSAR) is prone to phase noise overestimation and atmospheric interference, while point-pair-based methods (PSP-InSAR) often suffer from insufficient arc coverage due to high computational cost. To address these challenges, we propose EP-InSAR (Exhaustive Pairs InSAR), a time-series InSAR method based on dense arc linkage, suitable for heterogeneous urban scenarios with localized non-linear deformation. A sequential baseline strategy suppresses the deformation component in the spatiotemporal double-differenced phase, enabling efficient one-dimensional arc parameter estimation. Furthermore, a dual-threshold, iterative network expansion algorithm is introduced to approximate exhaustive identification of high-quality arcs, reducing reliance on amplitude-based priors and improving monitoring point density. Residual topographic and thermal signals are then removed from the interferogram stack to recover a refined deformation time series. Experiments on TerraSAR?X data over Shanghai demonstrate that EP-InSAR significantly increases point coverage, successfully recovers numerous high-quality points that violate conventional amplitude priors (e.g., 29.3% have ADI > 0.6), and yields a network of 910,000 points connected by 60 million high-quality arcs in a 3000×3000 window. This provides strong redundancy for parameter estimation and enables the detection of multiple localized deformation risks, corroborated by satellite historical imagery. The overall workflow has been preliminarily optimized for parallel processing, showing practical potential for large-scale structural health monitoring.