非线性卡尔曼滤波相位解缠中的关键问题研究

项目名称： 非线性卡尔曼滤波相位解缠中的关键问题研究

项目编号： No.40874001

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 刘国林

作者单位： 山东科技大学

项目金额： 40万元

中文摘要： 针对相位解缠中的非线性以及地形起伏大、噪声影响下相位解缠质量差等问题，从理论研究、模型建立、算法实现、实例分析、实验结果对比及验证等方面，围绕相位解缠中的关键理论问题进行了系统深入地研究。其主要内容包括：（1）针对地形起伏、噪声对相位解缠的影响问题，提出顾及地形坡度、误差影响的卡尔曼滤波相位解缠算法；（2）针对相位解缠的非线性问题，提出基于无味卡尔曼滤波、同伦算法、二维元胞自动机等的非线性相位解缠方法；（3）针对相位解缠中涉及的权重确定问题，提出基于频率置信度、相位导数变化图等的权重确定方法；（4）针对InSAR中涉及的其他关键数据处理问题，提出基于加权最小二乘的低相干SAR影像对精配准方法、基于相位点频率可信度的圆周期自适应滤波算法、基于三基线和最小的公共主影像优化选取方法以及基于小波相位分析的PS点识别与选取方法等；（5）采用不同卫星、不同分辨率、不同地形条件下的SAR数据进行了实验验证，取得良好的效果。 项目的开展，对提高地形起伏大、误差影响大等复杂地形条件下相位解缠的精度，进而提高数字高程模型的提取精度，特别是提高微小形变监测精度具有重要的理论意义和应用价值。

中文关键词： 相位解缠；卡尔曼滤波；非线性处理；权重

英文摘要： According to the problem in phase unwrapping that the nonlinear, and undulating terrain is large, the quality of phase unwrapping is poor under the influence of noise, the project is in the system and in-depth study around the key theoretical issues in the phase unwrapping, from theoretical research, modeling, algorithms, simulation and real examples, the comparison and verification of experimental results. Its main contents include: (1) It is proposed that the Kalman filter phase unwrapping algorithm that be taken into account for terrain slope and error affect, for the problem that undulating terrain and noise effect on phase unwrapping; (2) For the nonlinear problem of phase unwrapping, it is put forward that the nonlinear phase unwrapping methods based on the unscented Kalman filter, the Homotopy algorithm, two dimensional cellular automata and so on; (3) It is proposed that weight determining method based on frequency confidence and phase derivative change map, according to phase unwrapping involved in the problem of weight to determine; (4) For the other key data processing problems of InSAR,it is to put forward four methods.They are included that a new method for fine registration of a pair of low-coherent SAR images based on the weighted least square method, InSAR periodic median adaptive filtering algorithm based on the phase frequency credibility, a new method to optimize and select common master images based on the minimum sum of three baselines, and a new method of PS pixels identification and selection based on the wavelet phase analysis; (5) It is obtained good results under the experiment verification using different satellites, different resolution SAR data and different terrain conditions. By carrying out the project, to improve the accuracy of phase unwrapping under complex terrain conditions in the undulating terrain, error effect, thus increasing the extraction accuracy of Digital Elevation Models, in particular, to enhance the accuracy of the small deformation monitoring, has important theoretical significance and application value.

英文关键词： Phase unwraping；Kalman filtering；Nonlinear processing；Weight