地基SAR影像加权干涉叠加法及其应用

Weighted interferometric superposition method for ground-based SAR images and its application

为减弱影像噪声及影像失相干性的影响,提出地基SAR影像加权干涉叠加法。根据影像特征,提出热信噪比(TSNR)、相干系数、相位稳定性对地基SAR影像分组,并利用标准差标准化算法(Z-Score)对各像元点指标值进行定量处理,得到顾及3种影响因子后的影像综合权值矩阵,再进行加权叠加处理。以湘潭市建筑基坑240景影像及盾构下穿湘潭北站10758景影像为实验对象,对上述数据进行处理后并与测量机器人数据进行对比。结果表明,该算法在长时序地基SAR影像处理方面效果较显著,PS点数量增加了4102个,与测量机器人数据相对误差不超0.5 mm,且较均值叠加法在影像处理方面对干涉相位噪声更敏感,能有效减弱影像在时间维噪声,弱化失相干影像,获得形变信息的最优估计。

In order to attenuate the influence of image noise and image incoherence,this paper proposes a weighted interferometric superposition method for ground-based SAR images.According to the image characteristics,thermal signal-to-noise ratio(TSNR),coherence coefficient and phase stability are proposed to group the foundation SAR images,and the standard deviation standardization algorithm(Z-Score)is used to quantitatively process the index value of each image element point,so as to get the comprehensive weight matrix of the image after taking into account three kinds of influencing factors,and then carry out the weighted superposition processing.In this paper,we take the 240-view image of Xiangtan building pit and 10758-view image of shield structure penetrating through Xiangtan North Station as the experimental objects,process the above data and compare them with the data of the measuring robot.The results show that the algorithm is more effective in long time series foundation SAR image processing,the number of PS points increased by 4102,and the relative error with the measurement robot data is not more than 0.5 mm,and it is more sensitive to interfering phase noise than the mean superposition method in image processing,which can effectively attenuate the image noise in the time dimension,weaken the out-of-coherence image,and obtain the optimal estimation of deformation information.