Statistical SAR data characterization by bootstrapping

The characterization of SAR data is one of many important applications in radar remote sensing. In fact, it permits the performances of segmentation and classification algorithms to be improved. The authors introduce the bootstrap method to resolve the problem of correlation of data which limits the performance of statistical models to describe in a good way the illuminated targets. They demonstrate that the K-distribution is in good agreement with the experimental data when using the bootstrap method. The classical Pearson chart is extended by introducing the K-distribution and it is used to confirm the judicious choice of the K-distribution as a model to describe statistical data