A model for removal of speckle noise in SAR images (ALOS PALSAR)

Speckle noise is primarily due to the phase fluctuations of the electromagnetic return signals. Since inherent spatial-correlation characteristics of speckle in synthetic aperture radar (SAR) images are not exploited in existing multiplicative models for speckle noise, a speckle noise model is proposed here that provides a new framework for modelling and reducing the speckle noise. Both quantitative and qualitative criteria, including speckle reduction and texture preservation, are used to evaluate the performance of the proposed filter; one PALSAR (new Japanese sensor) image and a JERS-1 image are employed in the evaluation. The results showed that the proposed filter is slightly better than commonly used filters such as the Kuan, gamma, enhanced Lee, and enhanced Frost filters. The proposed filter can be used in different applications, including mapping and forestry biomass estimation. Furthermore, one of the benefits of the proposed filter is that it is independent of the threshold, which is required in most commonly used filters. The proposed filter was tested with SAR images of different sites in the northern forests of Iran.