基于PD算子的InSAR干涉图滤波研究

讨论了InSAR干涉图的滤波问题，提出了基于PD算子的干涉图滤波算法。算法首先对干涉图作分块处理，并假定其中各数据块上的相位值符合多项式模型，通过PD算子分别估计各数据块的模型系数，由此重构出近似干涉图。然后根据近似干涉图对原始干涉图进行‘解调’，降低原始干涉图的条纹密度，并使用滑窗作低通滤波处理。最后将滑窗处理后的干涉图与近似干涉图相乘得到滤波后的干涉图。为避免因分块处理而产生干涉相位图的马赛克效应，算法通过处理两组错位的干涉图分块数据，并对滤波后的两幅干涉相位图作加权处理，有效地降低相邻数据块间相位值的不连续性。对X＿SAR重复轨道数据产生的干涉图的处理结果表明，该算法能够对干涉图作很好的滤波处理，特别是在低信噪比区域，算法具有较突出的滤波性能。

InSAR Interferogram Filtering Based on PD Operator

InSAR interferogram filtering is discussed and a new filtering algorithm is presented based on the PD operator. The algorithm firstly divides the interferogram into the blocks. Assuming that the phases of each block are in accordance with multinomial model, the approximate interferogram is reconstituted after separately estimating the model coefficients of each block through PD operator. Secondly based on the approximate interferogram, the primitive interferogram is demodulated to decrease the density of interferometric fringe. So the low pass filter can be used to process the interferogram. Finally the filtered interferogram is multiplied with the approximate interferogram to resume its interferometric fringe. To weaking the mosaic effect on the interferometric phase image produced by the blocking process, the algorithm deals with two group of partly overlapped block data. And two filtered interferometric phase images are weighted to effectively improve the phase continuity of the adjacent data block. Experimental result of the interferograms generated by X -SAR repeat-pass data demonstrates the effectiveness of the algorithm, especially its outstanding ability to resume fringe in the low SNR areas of interferogram.