Contourlet transform based unsupervised change detection in synthetic aperture radar images

In this paper, intrinsic geometrical information of contourlet transform (CT) is used to propose unsupervised change detection over the data collected from directional subbands coefficients generated at various scale by using expectation maximization based parameter estimation. The unsupervised multi-scale and multi-directional changed detection method is presented by using same image of the synthetic aperture radar (SAR) images at the different time instant. The proposed approach exploits a CT-based multi-scale and multi-directional decomposition of the log-ratio image, which is obtained by taking the logarithm of the pixel ratio of two SAR image, aimed at achieving different scales of representation of the change signal. Experimental results carried out on multitemporal SAR images acquired by ERS-1 and JERS satellites confirm the effectiveness of the proposed approach.