Automatic target recognition using multipolar bistatic synthetic aperture radar images

Radar is a coherent sensor, which makes it suitable for the retrieval of the polarization information of the scattering process. Different algorithms have been proposed in the last few decades regarding how to analyze multipolar radar information. The extra information retrieved from the multipolar data has been shown to give more information regarding the physical features of the scatterer than the single polarization data. A few papers in the open literature have reported initial results regarding the use of monostatic multipolar data in radar-based automatic target recognition (ATR). However, a detailed report regarding the advantages of using multipolar data for ATR has not yet come in the open literature. Nor has there been any report on the use of multipolar data for bistatic radar-based ATR exercises. The work presented here deals with the task of using multipolar data for bistatic ATR. A group of algorithms have been proposed and analyzed for the task of multipolar bistatic synthetic aperture radar (SAR) ATR. Most of these algorithms are the contributions of the present work. For validating the algorithms, a simulated multipolar bistatic dataset has been used. It is also shown that using multipolar data not only improves the absolute performance of an ATR algorithm, it also makes it statistically more robust. Finally, the paper discusses a new algorithm to use multipolar data to avoid the decreased performance of an ATR algorithm that may arise from an uncalibrated bistatic multipolar radar system.