Combining Savitzky-Golay filters and median filters for reducing speckle noise in SAR images

This paper describes a new filter developed for reducing speckle noise in SAR images. The new filter, referred to as the Savitzky-Golay median hybrid filter, is a two-dimensional extension of the one-dimensional FIR-median hybrid filter. It operates by performing a series of two-dimensional polynomial least squares fitting, and defines its output as the median of the results obtained from such least squares fits. The polynomial least squares fitting performed in the new filter is efficiently implemented based on the simplified procedure developed in the Savitzky-Golay filter. The performance of the proposed filter is compared with that of the median filter, the Lee filter, and the Frost filter in reducing speckle noise of a synthetic image and of a real synthetic aperture radar (SAR) image.