IMAGE TEXTURE PRESERVATION IN SPECKLE NOISE SUPPRESSION
Ali Shamsoddini, John C. Trinder · 2010
Speckle noise which occurs due to the coherent imag ing system is the best known problem of SAR images and in turn, affects classification, change detection, biomass estimatio n and interpretation results. Several adaptive filt ering methods have been documented to deal with this issue, such as Kuan, L ee, MMSE and Frost filters. These filters do not co nsider the level of homogeneity in the intensity of the pixels. For thi s reason, they degrade the spatial resolution of im age and smooth details, while significantly decreasing the speckle noise level. T here are other filters such as Enhanced Lee and Gam ma Map that utilize the level of homogeneity, but they cannot adequately suppress speckle noise. Moreover, pixels whose coefficients of variation are near to maximum and minimum threshold values are not correctly filtered using these filters. In addition to th ese weaknesses, pixels surrounding a point scatterer are also treated as p oint scatterers due to shortcoming of the method of evaluating the coefficient of variation for differentiating between them and the point scatterer. We have developed a new method based on the homogeneity level for speckle noise suppression and simultaneously ed ge and feature preservation. Also, an algorithm has been proposed based on local statistical information to filter the pixels surrou nding point scatterers. The results show an improve ment in speckle reduction and texture preservation as well as reduction in the nu mber of unfiltered pixels.