条件随机场的多极化InSAR联合相位解缠算法  

A Combined Multi-polarimetric InSAR Phase Unwrapping Algorithm Based on Conditional Random Fields

作　　者：何楚[1,2] 石博[1] 蒋厚军[2] 罗元[1] 廖明生[2] 

机构地区：[1]武汉大学电子信息学院,湖北武汉430072 [2]武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079

出　　处：《测绘学报》2013年第6期838-845,共8页Acta Geodaetica et Cartographica Sinica

基　　金：国家973计划(2013CB733404);国家自然科学基金(60702041;41174120);中国博士后科学基金一等资助与特别资助

摘　　要：提出一种基于CRF模型的多极化InSAR联合相位解缠算法,用于在多极化方式的InSAR中选择最优通道的解缠相位计算高程。其中,缠绕相位的联合条件概率的倒数用作一元似然能量项,全变分对应的解缠相位的平滑度用作二元先验能量项,通过将两者统一的能量函数进行计算并优化,减少了相位解缠和单通道的噪声引起的误差。试验证明,该方法获得的高程精度优于最大似然估计高程。A Multi-polarization InSAR joint phase unwrapping algorithm is proposed. Multi-polarization tnSAR mode in searching the optimal channel variation based on model. In this approach, reciprocal of joint conditional probability of wrapping This method can be used in the conditional random field （CRF） phase is considered as unitary likelihood energy term, the smoothness of the unwrapping phase that corresponding to the total variation is considered as binary energy term. This method can decrease the error which is caused by the unwrapping phase and single channel noise through calculating and optimizing the two energy terms together. The experiments testify that our method can get a better elevation accuracy than maximum likelihood evaluation （MLE）.

关 键 词：INSAR 高程优化 全变分 条件概率 平滑度 

分 类 号：P237[天文地球—摄影测量与遥感]