Full polarimetric ground penetrating radar imaging using multiple measurement vectors model

Polarimetric ground penetrating radar (GPR) can enhance the imaging, detection and classification performance of GPR system by analyzing the polarization information of the echoes of buried targets. In this paper, a full polarimetric GPR imaging algorithm based on multiple measurement vectors (MMV) model is proposed. By utilizing the joint sparsity of polarimetric signals, the measurement data of different polarization channels are treated as MMV to account for interchannel correlations. Then the multiple polarimetric channel measurement data are processed jointly using multi-task Bayesian compressive sensing (MT-BCS) algorithm to produce several images of the investigated area for different polarization channels. Experimental results have shown that the proposed imaging algorithm is able to reconstruct the image of detected area with higher target-to-clutter ratio than the traditional single measurement vector (SMV) based imaging method.