中文摘要：

阐述合成孔径干涉雷达（InSAR）干涉条纹图的频率域和空间域滤波的算法及其优点与不足,着重分析复数空间自适应滤波的算法原理,提出2点改进：增加45°方向和135°方向的梯度计算及根据干涉图相干系数确定条纹边缘的尺度函数,从而进一步改善权系数。以模拟数据和我国香港及意大利Etna火山2地的真实干涉图数据为例,用本文方法与空间自适应滤波、Lee滤波、基于信噪比的Goldstein方法、最优化方向融合方法进行比较。实验结果表明：改进算法能够根据噪声强弱相应地控制滤波的强弱,既能更好地抑制相位噪声,又能较好地保持条纹边缘信息,模拟干涉图和香港干涉图残余点改善率分别提高18.6%和17.2%,表现出更强的局部自适应性。

英文摘要：

Algorithms and their advantages and disadvantages of filters for InSAR fringe images in frequence and space domains were described. The algorithm theory of the complex space domain adaptive filter was discussed. An improvement was made in two points, one was adding the gradient computation in 45° and 135° directions, and the other was the determination of scaling function for fringe edge based on interferogram coherence coefficient, thus modifying weight coefficients. According to the simulation data and actual interferograms of Hongkong and Etna volcano in Italy, the proposed method was compared with the complex space domain adaptive filter, Lee filter, Goldstein based on SNR （signal-noise-ratio）, and adaptive direction-dependent based on optimal integration. The experiment results demonstrate that this improved method has capability of tuning the filtering according to noise intensity. It performs well not only in noise reduction but also in preservation of the sharp sawtooth profile of fringe, with an improvement of 18.9% and 17.2% of residual points reduction in simulate and actual experiment respectively, showing stronger local adaptability.