Speckle reduction in synthetic-aperture-radar imagery.

Speckle appearing in synthetic-aperture-radar images degrades the information contained in these images. Speckle noise can be suppressed by adapted local processing techniques, permitting the definition of statistical parameters inside a small window centered on each pixel of the image. Two processing algorithms are examined; the first one uses the intensity as a variable, and the second one works on a homomorphic transformation of the image intensity. A statistical model for speckle noise that takes into account correlation in multilook imagery has been used to develop these processing algorithms. Several experimental results of processed Seasat-A syntheticaperture-radar images are discussed.