Clutter Learning Based LS Method for Buried Target Detection in GPR Images

  • Deniz Kumlu National Defense University/Naval Academy
  • Işın Erer Istanbul Technical University
Keywords: Least squares, Regularized least squares, Clutter learning, GPR, gprMax, RPCA


A regularized version of the least squares (LS) target detection method is combined with the subspace-based clutter learning for buried target detection in ground penetrating radar (GPR) images. The LS method is used to estimate the next A-scans from previously observed A-scans which are assumed to belong to the clutter component. Generally, A-scans used in the initial stage are accepted as target-free for the LS to work correctly. However, this is not guaranteed and if the first observed A-scan samples contain any target information, LS method will fail. In this paper, the clutter information is retrieved via robust principal component analysis (RPCA) as a preprocessing stage and used in the LS estimation of the actual A-scan. Thus for A-scans containing target information LS method will provide an increase in the estimation error indicating target presence at this location. Moreover, due to the regularization, the proposed method is more robust to noise caused by the irregularities of the soil.

How to Cite
D. Kumlu and I. Erer, “Clutter Learning Based LS Method for Buried Target Detection in GPR Images”, JAST, vol. 13, no. 2, pp. 217-225, Jul. 2020.