GPR image decomposition using two dimensional Singular Spectrum Analysis

Ground penetrating radar measurements can suffer from large amount of noise and clutter. Current methods, such as time gating and background averaging, mostly applied to remove reflections from air-ground interface do not perform well when removal of extraneous and very strong and non-uniform clutter signals originating from the objects in the surveyed area other than the target is needed. This work describes and evaluates performance of Singular Spectrum Analysis (SSA) and its multivariate derivatives for those tasks. Experimental GPR data using simple geometric shapes measured under laboratory conditions are used to demonstrate the effectiveness of proposed algorithm for these tasks.