多源SAR影像监测矿区建筑物三维位移场

Three-dimensional displacement field of buildings detection from multi-source SAR imagery

合成孔径雷达差分干涉(差分InSAR)技术仅能获取雷达视线向(LoS)一维形变信息.通过分析LoS向形变的三维空间合成模型,研究了融合多源合成孔径雷达(SAR)影像求解建筑物三维形变的模型与算法.该算法利用具有不同空间几何的SAR影像获取目标在不同LoS向的形变信息,联合求解目标三维形变场.结果表明:三维算法反演的垂直向位移(即下沉)与水准数据相互吻合,均方根误差为士10 mm;东西向水平位移场符合开采沉陷规律.由于卫星采用极轨飞行,使得三维算法对南北向位移不敏感,反演精度差;由三维位移场计算的倾斜值以及东西向水平变形与实测值间的均方根误差分别为±0.21,±0.06 mm/m.

Only one-dimensional displacement along radar line of sight （LOS） can be revealed using differential InSAR technique. The composition model of displacement along LoS was ana- lyzed firstly. And then a three-dimensional （3D） model and algorithm which fuses multi-source SAR images acquired with different geometric parameters in different satellite platforms to extract the 3D displacement field were presented in this paper. The experimental results show that the vertical displacement derived from 3D algorithm agrees on leveling data, and the root mean square error （RMSE） is ＋ 10 mm. The horizontal displacement field in the east-west direction accord with the law of surface movement caused by mining. And the precision of horizontal displacement in north-south direction is poor because all the SAR images are acquired from polar orbit satellite. The RMSE between the tilt calculated from vertical displacement, the horizontal deformation calculated from horizontal displacement and the leveling data are ±0.21 mm/m and ±0.06 mm/m, respectively.