Knowledge-aided object-oriented three-dimensional microwave imaging

For the past few years, researchers hold a strong interests on knowledge-aided object-oriented high-resolution microwave imaging field. In a novel framework of object-oriented microwave imaging using sparse prior based on the generalized reflectivity model is presented. The proposed methodology has several distinct advantages over existing methods in terms of the imaging quality: e.g., the generalized reflectivity model, which is a function of frequencies, viewing angles and polarization, allows us to consider more realistic interaction between the scene and the wavefield, and thus establishes a more accurate imaging model. Second, this novel methodology is a natural closed-loop iterative operation, which admits us to apply more prior knowledge (object-oriented feature knowledge) into the algorithm procedure resulting in greatly improving the imaging quality. In this paper, target classification stage is further introduced into the closed-loop iteration process, which bring a feedback to image processing stage for reducing the size of sample library and further improving the imaging quality. In this way, the pattern or local structure of imaged scene could be enhanced significantly. Another benefit introducing prior knowledge and target classification (or target identification) is the remarkable reduction of the transceiver elements. Selected simulation results are provided to demonstrate the state-of-art performance of the proposed methodology.