数字高程模型(Digital Elevation Model, DEM)描述的是地形表面高程信息，它在土木工程应用、景观设计与城市规划和军事目标建设等方面有着重要的作用。其中1:50000高精度DEM数据在土地利用现状调查、专题图制作、更新更小比例尺数据等方面有着广泛的应用。InSAR技术以任何天气条件下获取高精度大范围干涉影像的优势正在被越来越广泛的应用于高精度DEM生成。InSAR卫星在对地面目标进行成像观测时会产生主星位置和速度、斜距、基线长度以及非模糊相位等一系列的系统误差，而在干涉数据处理过程中各系统误差主要表现为干涉相位误差、基线测量误差和斜距测量误差，三者一起影响着干涉结果的精度。 针对干涉相位中的噪声误差，提出一种基于BEMD-Goldstein的干涉相位滤波处理方法。利用二维经验模态分解(bidimensional empirical mode decomposition，BEMD)算法对干涉图进行分解；采用分解后的结果对干涉图影像进行局部信噪比的估算，将局部信噪比作为Goldstein滤波新的滤波因子对分解后的噪声分量进行滤波处理；最终将滤波结果进行重构处理生成新的干涉图，用于后期数据处理。 针对山区等测绘困难区域获取地面控制点困难的问题，提出基于参考DEM进行干涉定标的方法。首先，基于斜距-多普勒模型将参考DEM模拟得到SAR图像坐标系下的模拟SAR影像，通过选取两景影像之间的同名点建立对应关系，获取SAR影像上对应点的高程信息；然后，将基线参数进行三维坐标分解对干涉定标模型进行优化，引入牛顿下山迭代法对优化后的定标方程进行解算处理，提高敏感度方程的解算速度与精度。 使用模拟和真实的干涉相位数据对提出的滤波算法进行验证分析。结果表明，算法在去除相位噪声和细节信息保持方面要优于其他滤波方法，可以为InSAR技术获取DEM提供更精确的干涉相位。同时，采用陕西省阎良地区的TanDEM-X数据对基线定标算法进行了验证，并用实测控制点和参考DEM对监测结果进行了对比验证。验证结果表明，检查点的高程均方根误差为1.95m，而整个研究区域的均方根误差为2.03m，满足1:50000比例尺测图精度的需求。此外，论文还使用山西太原以及西藏双湖地区的TanDEM-X影像进行了实验。实验表明，论文方法在山地和高山地等地形复杂地区都可以获得高精度DEM结果。论文对将InSAR技术工程化应用于DEM生产具有重要的参考价值。

Digital elevation model (DEM)is a description of the terrain surface elevation information. It plays an important role in the civil engineering applications, the city landscaping and planning,the military construction, etc. Particularly, the 1:50000 high precision DEM data has a wide range of applications in the current situation of land use survey, thematic map production, update the smaller scale DEM data and other aspects.InSAR methodhas been widely applied in high precision DEM acquisition due to the advantages ofwide monitoring range, high monitoring accuracy, all weather monitoring and other advantages.The research shows that Radar satellite includes the system errors of position and velocity, range distance measurement, baseline measurement and no fuzzy phase in the imaging observation of ground targets. These system errors will show asinterferometric phase error and baseline error during the InSAR processing which affecting the accuracy of the final results. According to the interferometric phase error, The BEMD-Goldstein interferometric filtering algorithm was proposed in this paper. At First, the bidimensional empirical mode decomposition (BEMD) algorithm was used for interference graph decomposition; then,the localsignal-to-noise ratio of the interferometric phasewas estimated by decomposition results, whichwill be the new factor in the Goldstein filtering algorithm for the decomposition filtering. Finally, the filtering results were reconstructed for a new interferogram which was used for the last processing. Secondly, aiming at the problem of obtaining ground control point in mapping difficult areas, a new method of interferometric calibration based on reference DEM was proposed. Based on theslant rangeand Doppler model for reference DEM, realtionshipwas established between SAR and simulation SAR for obtaining the corresponding elevation information in SAR image. Thenthe elevation was used for the baseline calibration processing.In order to improve the accuracy of the sensitivity equation, the method of Newton descent was used to optimize the existing calibration models. Finally, the filtering algorithm was verified by simulated and measured images. The results show that the proposed algorithm can effectively maintainthe interferometricedge details while suppressing the phase noise, and providing a better filtering interferometricfor the acquisition of high precision DEM. At the same time, the TanDEM-X data of YanLiang area in Shanxi province was used to verify the interferometriccalibration algorithm, then the results were provedbysome measured control points and the referenced DEM. The results show that the root mean square error of the check point was 1.95m, and the root mean square error with the referencedDEM was 2.03m, which meets the requirements of 1:50000 topographic mapping whichprovedthe algorithm's reliabilityproposed. In addition, TanDEM-X images of Taiyuan in Shanxi Province and Shuanghu area in Tibet were also used. The experiments show that the proposed algorithm canprovidehigh precision DEM resultsnot only in mountains but also in high-mountain area and other complex terrain areas. The method has an important value for the engineering application of InSAR technology.