Método de log-cumulantes em processamento de imagens SAR

Abstract

Image segmentation can be applied to a broad class of different problems. However, it is not usually a simple task for Synthetic Aperture Radar (SAR) images due to the presence of speckle. Given the importance of SAR images in remote sensing problems, this thesis introduces a general and simple methodology to achieve SAR image segmentation by using the estimated roughness parameters of SAR data modeled by $G_I^0$ and $G_A^0$ distributions, instead of directly processing the speckled images. In this paper, we adopted the log-cumulants method for the roughness parameter estimation. The performance evaluation of the results was attained in terms of the Error of Segmentation and Cross-Region Fitting measures for synthetic and real SAR images, respectively. Regarding synthetic images, we performed Monte Carlo experiments which confirmed the suitability of SAR image segmentation by means of roughness parameters. The results showed that the methodology provides a feasible input to SAR image segmentation algorithms which also include thresholding based methods. Actually, the proposed approach accomplished satisfactory results for the most critical case study, the single-look images, which are markedly affected by speckle. The application of the log-cumulants method to synthetic aperture radar data processing encompasses parameter estimation of probability density functions. The good estimation performance of this method has fostered researches in SAR data modeling and processing with the $G_I^0$ distribution. Numerical methods are usually applied to estimate parameters of the $G_I^0$ distribution by log-cumulants and therefore they can result in high computational cost. Here, we propose a fast log-cumulants approach for SAR data modeled by the $G_I^0$ distribution. Experimental tests were carried out on sets of simulated and real SAR data.