Analysis of SAR ImagesDe-speckling using a Bilateral filter and Feed Forward Neural Networks

Abstract

Speckle noise reduces the quality and nature of SAR imageries and diminishes the performance of SAR image processing. Thus, the multiplicative noise must be stifled before processing the image utilizing different image handling systems. Even though, there are number of speckle noise reduction techniques are available, all have its own merits and demerits. Therefore, noise reduction is still a major impediment in SAR image processing. In this paper, the speckle noise is reduced by using neural Network followed by the Bilateral Filter. This paper also presents the comparative analysis of two layered FFBPNN, TLFFBPNN and FLFFBPNN for speckle noise reduction of SAR images. Upon comparisons, it could be concluded that, TLFFBPNN de-speckling method provides good visual effects of SN reduction with better similarity and edging conservation metrics.