Non-subsampled contourlets based Synthetic Aperture Radar images segmentation

Abstract

It is well known that the Synthetic Aperture Radar(SAR) images are abundant of directional and texture information, which is very useful for segmentation. Contourlet is a geometric multiscale tool that is based on multiscale filters and directional filter banks. It not only inherits the multiscale characteristics of dimensionality-inseparable wavelets, but also has the flexible multi-directional characteristic. In this paper, we developed a new non-subsampled contourlet transform (NSCT) and gray level co-occurrence matrix (GLCM) based image segmentation method for SAR image segmentation. For the redundant and shift-invariant property of the NSCT, and the statistical texture features extracted by GLCM, the proposed method can present accurate segmentation result for SAR images.