New Robust Sparse Convolutional Coding Inversion Algorithm for Ground Penetrating Radar Images

In this paper, we propose two algorithms to enhance the interpretability of hyperbola in B-scans obtained with a Ground Penetrating Radar (GPR). These hyperbolas are responses buried objects or cavities. To correctly detect and classify them, denoising is typically necessary for GPR images as signal-to-noise ratio low, various interfaces naturally present earth have strong response. Both based on sparse convolutional coding model plus low rank component. It solved through an Alternating Direction Method Multipliers (ADMM) framework. order take into account presence outliers artifacts caused by acquisition, second algorithm Huber norm instead classic L 2 -norm. tested real dataset labeled geophysicists. The results show efficiency approach, particular robustness algorithm.