雷达干涉测量图像的梯度自适应光滑子函数滤波法

提出了基于梯度的自适应光滑子函数滤波的新方法。对于干涉图中的点，通过梯度分为噪声点和非噪声点。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，非噪声点处不进行平滑处理。通过这种方法有效抑制了斑点噪声，又同时做到使干涉图相位偏差较少和较好地保留了边缘信息。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。

APPLICATION OF ADAPTIVE SMOOTHING FUNCTION FILTERING BASEED ON GRADIENT IN InSAR IMAGE

A new filtering algorithm smoothing function based on gradient was proposed. Based on gradient, noise points are identified. This algorithm is based on the spatial curved surface fitting, and then the surface is processed through smoothing function. The good pixels are unsmooth. The results of experiment, using the data which come from ALOS data of Beijing, show that the smoothing function filer has effectiveness and feasibility. The RMS (Root mean square error), PSNR (Peak signal-to-noise ratio), EPI(Edge preserve index) and PSD (Phase standard deviation) suggest that the algorithm can reduce the speckle noise and keep the edge information.