质量图指导的改进的基于CMRF模型的自适应InSAR相位滤波

Quality Guide Improved CMRF Model-based Adaptive In SAR Phase Filter

干涉相位图滤波是In SAR数据处理中的关键步骤,滤波效果的好坏直接影响后续相位解缠的准确程度和难易程度。一种基于复值马尔可夫随机场(complex-valued markov random field,CMRF)模型的干涉图滤波算法能够有效的保持边缘信息并降低残差点的数目。该算法的滤波效果依赖于模型参数估计的准确程度,而且滤波后的相位图不够平滑,仍然存在很多噪点。针对上述两个问题提出了质量图指导的基于CMRF模型的自适应滤波算法。首先,在CMRF模型参数估计中利用不规则窗使得每个待更新像素都有一个符合自身局部统计特征的更准确的模型参数。其次,引入了质量图来指导干涉图各个部分的平滑程度。仿真和实测数据的结果验证了改进算法的有效性。

InSAR phase filtering is a key step in the InSAR data processing. It affects the accuracy and com- plexity of the phase unwrapping process. An adaptive phase filter based on complex-valued markov random field （CMRF） can preserve phase jumps and reduce residues. It uses the estimated CMRF model parameter to update the residues and their neighbors. Therefore, the filtering result depends on the accuracy of the model parameter es- timation. Moreover, the filtered phase map is not smooth. To solve the two problems above, firstly, the irregular window is used to calculate each pixel＇ s model parameter. Then, the quality map is introduced to guide the smooth degree of the filtering result. The simulated and real data experiments are performed to validate this method.