17 December 1996 Adaptive vectorial speckle filtering in SAR images based on fuzzy clustering criteria
Author Affiliations +
This paper deals with adaptive vector filtering of speckling in SAR images. The proposed method is based on the ordering of the vector data. The vector order is obtained by an Euclidean distance calculated from the center of a set of vectors. The local variation coefficient has been introduced for filter adaptivity. This coefficient is a reflection of the generalization of local scalar variation towards a case of a vectorial variation. For this purpose we use the fuzzy cluster analysis domain.In each large window, we decompose data into two groups using the fuzzy center mobile algorithm. The Euclidian distance between the two group centers provides good information about local variation in the window. In a homogeneous area this distance is minimal whereas it is considerable in the heterogeneous area. The global distribution of this distance coming from all windows of the image determine the threshold value. According to the local variation value, the vector will be treated in two different ways. If the local variation is less than the threshold value then we use a classic mean filtering. Otherwise, we replace the vector of interest by another ordered vector. Its choice depends on the local variation value.
© (1996) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Ali Saad, Ali Saad, Safwan El Assad, Safwan El Assad, Dominique Barba, Dominique Barba, } "Adaptive vectorial speckle filtering in SAR images based on fuzzy clustering criteria", Proc. SPIE 2958, Microwave Sensing and Synthetic Aperture Radar, (17 December 1996); doi: 10.1117/12.262690; https://doi.org/10.1117/12.262690


Back to Top