基于幅相不一致准则的建筑物SAR层析成像

SAR Tomography Imaging for Buildings Using an Inconsistency Criterion for Amplitude and Phase

传统的谱分析和压缩感知(CS)等层析合成孔径雷达(TomoSAR)成像技术由于解斜处理需要估计得到精确的垂直雷达视线(PLOS)方向的垂直有效基线。为了避免此操作该文采用沿PLOS向进行搜索的空域波束形成(BF)方法进行层析聚焦。由于高分辨率SAR图像中城区建筑物结构复杂,不同航过SAR图像间存在观测视角差异,并且存在相干斑噪声影响,SAR图像中的所有同源点不能在相同的像素点同时进行精确地配准。为了从幅度和相位两个方面找出BF聚焦时最相关的像素点,提出了一种联合参考像素窗口中邻域像素点的幅度和相位来提取目标像素点的不一致性准则。根据不一致性准则的最小化提取出相应的同名像素点,以实现对层析成像的精确聚焦。利用仿真数据和高分辨X波段重复航过机载SAR系统录取的实测数据进行实验。实验结果中,利用传统方法聚焦得到的散射轮廓峰值位置在15.63 m而该文所提方法得到峰值位置在16.88 m,峰值位置更加接近建筑实际高度18 m。表明该方法能有效提高散射体在PLOS方向的聚焦能量,并精确提取建筑物的3维轮廓。

An indispensable step in the imaging of Tomographic Synthetic Aperture Radar(TomoSAR)in spectral analysis or Compressive Sensing(CS)technology is estimating the perpendicular baselines in the Perpendicular Line Of Sight(PLOS)in deramping operations.To avoid this procedure,we introduce a Beam Forming(BF)method in the spatial domain that scans for TomoSAR focusing in the PLOS direction.Because of the sophisticated structure of buildings in urban areas,multipass high-resolution SAR images suffer from discrepancies in the look and incidence angles as well as speckle noise.As a result,it is challenging to precisely coregister all the homologous points in the identical pixels of multipass SAR images.To identify the most relevant pixels with respect to both amplitude and phase when BF imaging is implemented,we propose an inconsistency criterion for specific pixels using the joint phase and amplitude of pixels in a window.By minimizing the inconsistency criterion,homologous points with high accuracy can be identified by focusing on TomoSAR imaging.We used simulation and real data from a multipass X-band airborne TomoSAR system in China to test the effectiveness of the proposed method.Experimental results show that the peak of the reflectivity profile via conventional tomographic imaging is about 15.63 m,whereas that by the proposed method is 16.88 m,which is very close to the actual height of the 18 m building.The results demonstrate the feasibility of improving the focusing power of scatterers in the PLOS direction and extracting the threedimensional outliers of buildings.