A new adaptive filtering algorithm using smoothing function based on phase gradient is proposed.The noise points in interferogram can be judged out through gradient of phases.At the points InSAR image window is fitted to quadric surface,which can be processed by smoothing function.This method is compared with other common filtering methods using the genuine ALOS data of Beijing area and the feasibility and effectiveness of which are verified by quantitative indicators such as RMS,PSNR,EPI and PSD,which suggest that it reduces the speckle noises effectively and keep the edge information well.The phase deviation of interference in figure is less.%提出基于梯度的自适应光滑子函数滤波的新方法。通过相位梯度判断出干涉图中的噪声点。在噪声点处，将包含噪声点的 InSAR 图像窗口拟合为二次曲面，然后用光滑子函数对拟合曲面进行平滑去噪处理。利用北京地区的真实ALOS数据进行滤波试验，并与几种常用的滤波方法进行比较和分析。通过均方根误差、峰值信噪比、边缘保持指数和标准相位偏差等量化指标，验证了基于梯度的自适应光滑子函数滤波方法的可行性与有效性。此方法可有效抑制斑点噪声，较好地保留边缘信息并且干涉图相位偏差较少。
展开▼
