Speckle Reduction in SAR Images by Using Homogeneity NeighShrink

Abstract

Synthetic Aperture Radar (SAR) image suffers from severe artifacts caused by speckle noise which is multiplicative in nature. Some of the adaptive filters such as the Lee filter, the Frost filter and the Kuan filter are the well known speckle filters. These filters adapt the filter coefficients based on the pixels within a fixed moving window. Though it removes speckle noise well in the homogeneous regions, it leaves noise in the heterogeneous areas to preserve the edges and fine details or smoothes the edges to remove the noise in this area. In order to reduce the speckle noise with the preservation of edges, a new method based on wavelet analysis and adaptive mean filtering is proposed in this paper. A comparative study with other methods show that proposed method is better in preserving edges and fine details while reducing the speckle noise.