Speckle reduction of SAR images using adaptive regularized least square support vector machines
Daiqiang Peng, Jinwen Tian, Jian Liu · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2007
Synthetic aperture radar (SAR) images are inherently affected by multiplicative speckle noise, which is due to the coherent nature of the scattering phenomenon. This paper proposed an adaptive regularized approach to reduce SAR image speckle based on least squares support vector machines (LS-SVM). Generally, SAR images are comprised of multiple features of different spatial scales, and there is typically a trade-off between speckle removal and detail preservation. A natural approach to partially alleviate this problem is to use spatial adaptive regularization parameter on the use of regularized procedure. Here, each pixel has its own associated regularization parameter in this paper, instead of choosing a global regularization parameter. Experimental results show that our approach has a good performance on the speckle reduction without destruction of important SAR image details.