Spatially adaptive polarimetric image despeckling using bandelet transform

Abstract

Polarimetric Synthetic Aperture Radar (PolSAR) imaging extended SAR applications by exploring the polarimetric properties of the target scatterers. Similar to SAR images, PolSAR images are prone to multiplicative speckle noise due to its coherent imaging mechanism. Hence, despeckling is an essential procedure for effectively utilizing the PolSAR images for remote sensing applications. Latest advances in PolSAR filtering techniques have shown the trend of using spatial domain techniques based on multiplicative speckle noise model which impairs the retrieved image by oversmoothing and absence of image finer details. More efficient despeckling algorithm using bandelet thresholding is proposed in this paper. Since the bandelet transform is computationally expensive, a spatially adaptive technique using directional variance is proposed. Performance evaluation with both airborne and spaceborne radar images has shown the approached technique is much efficient related to speckle reduction and feature preservation compared to the state of art techniques.

Introduction to the special issue on deep learning for remote sensing environments

“…The article about "Spatially Adaptive Polarimetric Image Despeckling using Bandelet Transform" (Thankachan et al, 2019) proposes about an efficient despeckling algorithm using bandelet thresholding in this paper. The Performance evaluation with both airborne and spaceborne radar images has shown the approached technique is much efficient related to speckle reduction and feature preservation compared to the state of art techniques.…”