Skip to Main Content
This paper investigates the effect of apodization on the statistical properties of synthetic aperture radar (SAR) images and its impact on the capability of extracting information from homogeneous regions of apodized SAR images. The statistical model for the pixel complex amplitude of the apodized image is derived in terms of both probability density function and statistical moments. Knowledge of the statistical properties is then used to develop appropriate schemes for parameter estimation and supervised classification of homogeneous regions with different radar cross sections in apodized SAR images. The performance analysis shows that the new techniques (properly derived for the apodized case) provide information extraction capabilities only slightly worse than those provided by the conventional techniques applied to the nonapodized case. This allows us to conclude that the use of nonlinear apodization yields sidelobe level reduction and main lobe resolution preservation that can be traded with the small losses above. A full characterization of the estimation and classification performance of these new techniques shows that nonlinear apodization globally introduces a performance degradation comparable to a reduction of the number of looks of a factor of 1.455 for a homogeneous region.