Complex coherence estimation based on adaptive refined lee filter

Abstract

干涉相干性包含了极化SAR干涉测量(PolInSAR)中的极化和干涉信息,滑动窗口大小和滤波方法是准确估计干涉相干性的前提。本文以精致极化Lee滤波为基础,根据边缘检测和邻域相干性高低区分同质区域和异质区域,建立了自适应精致极化Lee滤波方法并估计相干性矩阵和干涉相干性。自适应精致极化Lee滤波能够根据边缘检测信息和邻域相干性高低程度调整滑动窗口大小并选择合适的估计方法,不仅提高了抑制斑点噪声的能力,而且保持了图像的边缘信息,有利于提高干涉相干性的估计精度。试验结果表明,该方法有效抑制了斑点噪声,较好地保持了SAR图像的边缘信息,有利于提高植被参数反演的精度。 Polarimetric synthetic aperture radar interferometry (PolInSAR) data procedures and their application are based on the estimation of polarimetric complex coherence, which are influenced by size of windows and filter methods. In this paper, the adaptive refined Lee filter, which based on traditional refined Lee filter, is used to estimate the interferometric coherence. The size of filter window is changed by the correlation coefficient between the central sub window and the neighboring sub window. Correlation coefficient which is larger than the threshold value means to the homogeneous pixels in the selected window, and then boxcar filter is chosen to estimate complex coherence. When maximum of correlation coefficient in difference windows sizes is smaller than the threshold value, the refined Lee filter is used to estimate the complex coherence. The efficiency and advantage of the new algorithm are demonstrated with E-SAR data sets. The results show that the influence of speckle noise and edge information is improved; more accurate complex coherence estimated by selected window size and selected pixels increase the accurate of forest parameters inversion.