Polarimetric SAR speckle filtering for high-resolution SAR images using RADARSAT-2 POLSAR SLC data

In this paper, an algorithm in polarimetric synthetic aperture radar (POLSAR) speckle filtering is proposed, which is more suitable for high-resolution SAR images. Firstly, the filter applies a method based on scattering model to retain strong point targets and avoid their effect on other pixels around. Then it uses statistic characteristic, improved edges and lines detector, as well as adaptive window size regulation to smooth speckle in homogeneous areas and preserve details in non-homogeneous areas. At last, this algorithm follows Lee´s principle when filtering the covariance matrix to preserve polarimetric properties and avoid crosstalk between polarization channels. To verify this algorithm´s validity, RADARSAT-2 C-Band POLSAR SLC data are used. Results show that the proposed filter performs well in speckle reduction, strong point target and structural feature retention and polarimetric properties preservation, compared with Boxcar, Refined Lee POLSAR Speckle Filter and 5MBSF.