A Novel Method for Polarimetric Sar Image Speckle Filtering and Edge Detection

Speckle filtering and edge detection are fundamental techniques for polarimetric synthetic aperture radar (PolSAR) image analysis. In this paper, based on complex Wishart distribution and the similarity test of the covariance matrix, a completely adaptive speckle filtering is proposed. Meanwhile, a new edge indicator named similar pixel number is developed. This indicator shows great sensitivity to determine the edge pixels and a new edge detection method is proposed. The novelty of this edge detector is that the edge can be detected without mono-edge assumption and the edge orientation can be extracted automatically without any window configurations and searching steps. The efficiency of the proposed speckle filter and edge detection scheme are demonstrated by ALOS/PALSAR data set.