Optimized Reconstruction of GPR Target Signature Using Evolving Sparse Representation

Penulis: Budiman Putra Asma'ur Rohman S.T.,Masahiko Nishimoto
In this paper, a new approach to signal processing based on sparse representation technique is presented to reconstruct the target signature of ultra-wideband pulse ground penetrating radar. The proposed method employs a well-known matching pursuit which is sequentially optimized by the evolutionary algorithm with a certain technique. Though the decomposition optimization by evolutionary algorithms has been proposed in some prior research, this study will extend the approach by taking a different path to explore further and optimize it. The simulated case has been taken in the free space in order to focus on the proposed method investigation. After determination of the time gating of target object response, the signal is decomposed by matching pursuit. In each decomposition step, the decomposed atom signals evolve themselves for refining their waveform parameters constrained by prior matching pursuit results in both lower and upper limit with a half offset. The objective function of the optimization is the sum squared error value. The simulation results show the superiority of the proposed method in the signal reconstruction which is promising enough to be applied for wide applications such as classiflcation and identification of the buried object. Then, the comparison with the existing methods was also conducted to confirm the performance of this method.

PIERS 2018 Toyama

No. Arsip : LIPI-20180712