The relationship between backprojection and best linear unbiased estimation in synthetic-aperture radar imaging (Q326395): Difference between revisions

The paper is devoted to investigate the relationship between two different statistical methods used in imaging and estimation problems: backprojection in synthetic aperture radar (SAR) imaging versus best linear unbiased estimation (BLUE), a core method in classical statistical science. The main goal in radar imaging is to estimate the reflectivity function of an object of interest which is present in a scene viewed by the radar. The radar emits electromagnetic pulses and collects measurements of the field scattered off the object and its surroundings. Then these collected scattered field measurements are used to estimate the unknown reflectivity function, in order to reconstruct the image of the scene of interest. The article starts with the derivation of the expression for SAR data from Maxwell's equations. Further, the investigations are carried out under the following assumptions: the image over the entire ground plane is flat, the data collection manifold is unlimited (i.e., infinite bandwidth and synthetic aperture), and white (i.i.d) additive noise has a zero-mean. Under these assumptions it is shown that the estimate of the reflectivity (calculated using BLUE) and the reconstructed image (calculated using filtered backprojection) are the same when a criterion from microlocal analysis is applied in order to define the optimal backprojection filter. Finally, it is worth mentioning that the first two assumptions are not realistic to any SAR imaging scenario but simplifies the analysis, and that ``it is impossible to obtain a truly unbiased estimator by using this backprojection/BLUE estimate in the case of a true SAR imaging scenario'', as noted by the author.