Multidimensional Speckle Noise Model

Abstract

One of the main problems of SAR imagery is the presence of speckle noise, originated by the inherent coherent nature of this type of systems. For one-dimensional SAR systems it has been demonstrated that speckle can be considered as a multiplicative noise term. Nevertheless, this simple model cannot be exported when multidimensional SAR imagery is addressed. This paper is devoted to present the latest advances into the definition of a multidimensional speckle noise model which does not depend on the data dimensionality. Speckle noise may be modeled by multiplicative and additive noise sources, whose combination is determined by the data's correlation structure. The validity of the proposed model is demonstrated by its application to a real L-band multidimensional SAR dataset acquired by the German ESAR sensor.

Keywords

• multidimensional SAR imagery
• speckle noise
• noise modeling