DESPECKLING OF POLSAR IMAGES

Abstract

Polarimetric synthetic aperture radar is an extension of Synthetic Aperture Radar (SAR) which is used to create 2D or 3D images in different polarization. It uses motion of radar antenna to the target object to get spatial properties better than other radars. PolSAR image is represented by a scattering matrix because it can send and receive in two polarizations that are horizontal and vertical. PolSAR led to many applications in disaster management, land cover mapping, forest mapping and agriculture. PolSAR is widely used in remote sensing applications. The main problem that arise after extraction of image is noise which degrades the quality of information that can be extracted from the image. This noise can be additive white Gaussian noise (AWGN) or granular multiplicative noise i.e. speckle noise. Performance of PolSAR is mainly affected by speckle noise. Speckle noise is a granular noise that inherently exists in PolSAR and degrades the quality of the active radar (SAR). Filtering techniques are commonly used to remove the effects of the speckle noise. The main goal of existing speckle filtering algorithms is to reduce the speckle while preserving the information. Some techniques are implemented and their comparative results are discussed. There comparative results of power are discussed. During despeckling with the removal of noise, classification accuracy is also affected. Classification accuracy of Lee Sigma filter and Refined Lee filter is discussed. Along with removal of noise, edges which describe the boundaries of the image should also be preserved. Edge detection is carried out by using canny edge detection with adaptive algorithm. Comparative study of canny edge detection with that of Sobel filter is performed and it has been observed that canny edge detection is better than Sobel filter.