LAUSR

Speckle filtering in synthetic aperture radar polarimetry (PolSAR) is essential for the extraction of significant information. In this study, the authors introduced a hybrid iterative filtering scheme. The proposed iterative filter is initialized by a polarimetric filter ensuring a high speckle reduction level. Then, to enhance the spatial details, the iterative filter is applied for few iterations. The key parameter $b_{k}$ which traduced the variability of the pixels has been studied. An expression that takes into account the following important numerical considerations is proposed. 1) The variability of the pixel is measured by coefficient variation (CV) rather than the variance. 2) The statistics are computed using non-local neighborhood instead of local square one. 3) A new normalizing function (i. e. hyperbolic tangent) is implemented. 4) The dynamic of range of $b_{k}$ is increased by multiplying the CVs of the filtered and the originals images. Comparison with various state of the art PolSAR filters demonstrated the effectiveness of the proposed iterative method. Simulated, one-look and multilook real PolSAR data were used for validation.