SYNTHETIC APERTURE RADAR IMAGE ENHANCEMENT USING PARTICLE FILTERS

In this paper, we propose a novel approach to reduce the noise in Synthetic Aperture Radar (SAR) images using particle filters. Interpretation of SAR images is a difficult problem, since they are contaminated with a multiplicative noise, which is known as the “Speckle Noise”. In literature, the general approach for removing the speckle is to use the local statistics, which are computed in a square window. Here, we propose to use particle filters, which is a sequential Bayesian technique. The proposed method also uses the local statistics to denoise the images. Since this is a Bayesian approach, the computed statistics of the window can be exploited as a priori information. Moreover, particle filters are sequential methods, which are more appropriate to handle the heterogeneous structure of the image. Computer simulations show that the proposed method provides better edge-preserving results with satisfactory speckle removal, when compared to the results obtained by Gamma Maximum a posteriori (MAP) filter.