Low complexity clutter removal in GPR images via lattice filters

A low complexity ground-penetrating radar (GPR) clutter removal method based on lattice filter (LF) decomposition is proposed. The GPR image is decomposed into several levels by LF where each level contains approximation and directional subbands. Due to its horizontal structure, the clutter is contained in a horizontal subband, while vertical and diagonal subbands represent the target information. Although the proposed approach seems similar to the classical undecimated wavelet decomposition, the whole process is accomplished in the spatial domain by simple mathematical operations and time delays, thus it has less complexity compared to the wavelet transform which needs the use of fast Fourier transform operations. Moreover, due to the recursive nature of the LFs, the proposed method is more robust to noise. It presents a better detection performance as indicated by the quantitative results. Our method is also compared with the conventional subspace–based methods for both simulated and real datasets, and visual and quantitative results validate the superiority of our proposed method.