Stepped Frequency Continuous Wave Ground Penetrating Radar Imaging Algorithm Based on Multi-task Bayesian Compressive Sensing

Abstract

Aiming at the joint sparsity of target imaging space of SFCW-GPR, the idea of multi-task learning is introduced into the Bayesian compressive sensing reconstruction algorithm, multi-task Bayesian compressive sensing(MT-BCS) algorithm is proposed in the paper. A common prior hierarchical Bayesian model is used for different tasks, MT-BCS algorithm can fully exploit the statistical characteristics and structural information of radar echo data and recover original signals from far fewer random samples. The simulation results show that the reconstruction accuracy of MT-BCS algorithm is better than BCS algorithm under the same conditions.