InSAR干涉图滤波方法研究

Research on Filtering Methods of InSAR Interferogram

探讨多视滤波法、中值滤波法、基于梯度的自适应滤波、additive滤波法四种抑制干涉图噪声的滤波方法。多视滤波法平滑了影像数据,是以牺牲空间分辨率为代价的,通常这种滤波处理应用在从两个单视影像获得的复数影像处理中;作为一种传统的抑制噪声方法,中值滤波技术实质上是一种非线性信号处理技术,它假设噪声具有极端的数值,即在所定义的平滑模板内为最(较)大值或最(较)小值,因此它会使得干涉图丢失一些信息;基于梯度的自适应法是基于梯度的一种中值滤波,它使得边缘更加清晰,该方法可与中值滤波联合使用;Addtive滤波法强调根据局部噪声状况和使用方向平行窗口得到的滤波噪声边缘来自适应的滤除噪声,对于局部噪声状况由关联图来决定。这种方法尤其对高关联的干涉图最为可取。

Four noise filtering methods are studied in this paper.Multi-look filter processing is implemented by averaging neighboring pixels in complex images.It is very useful,but at the cost of spatial resolution.In order to get better result,range and azimuth critical look number should be estimated according to the calculation formulas.Usually,this filter processing is applied in complex images obtained from two single-look images. Then,interferogram is formed and it is clearer.As a traditional noise restricting method,medium filter is also useful in InSAR.It supposes phase noise to be at the extreme value.But,if it is not,error will occur.So,this filter will lose some information in interferogram.Gradient-based adaptive filter is a mean filter based on the gradient and it makes the fringes clearer.If the pixel is in the edge,it will smooth down less,else more.Usually,some noise looks as bur.So,it will be preserved as fringe and unable be removed.Therefore,this filter is used to remove the speckle noise.When it combines with medium filter,it can be performed effectively.Addtive filter emphasizes filtering noise adaptively according to the local noise level and filtering noise fringes using directionally corresponding windows.The local noise level is determined with the help of coherence map.This method is available,especially to the interferogram with higher coherence.