Optimization of Speckle Filtering Techniques for ERS-1 Images over Argentina

摘要

SAR images provide detailed information about the target surface, though it isobscured by the well-known speckle effect. One alternative of speckle reduction is by filtering the image, but this demands some assumptions about the speckle behavior. However, it is important to take into account that noise reduction filters tend to blur the edges and sharp features of the images, since they always introduce some kind of low pass effect. Several filters have been developed for the speckle reduction, and most of them are of the form of a window that slides over the image. But there is no automatic process to determine the window size, a relevant parameter that depends, among other things, on the features of the studied area. Given a filter, it can be applied to the SAR image for different windows then it is possible to estimate the so-called speckle index as a measure that provides the best speckle reduction in accordance to the degree of heterogeneity of the region under study. In this paper the type of filter which better reduces the speckle in homogeneous and heterogeneous regions is determined. Moreover, it is possible to draw conclusions about the optimal window size used when the filter is applied.