Visible to the public Compressive Sampling Stepped Frequency GPR Using Probabilistic Structured Sparsity Models

TitleCompressive Sampling Stepped Frequency GPR Using Probabilistic Structured Sparsity Models
Publication TypeConference Paper
Year of Publication2021
AuthorsKafedziski, Venceslav
Conference Name2021 15th International Conference on Advanced Technologies, Systems and Services in Telecommunications (℡SIKS)
Keywordscomposability, compressive sampling, Estimation, Ground penetrating radar, Image coding, land mine detection, Landmine detection, Markov chain, Markov random field, Probabilistic logic, probabilistic model, pubcrawl, radar antennas, radar imaging, resilience, Resiliency, stepped frequency radar
AbstractWe investigate a compressive sampling (CS) stepped frequency ground penetrating radar for detection of underground objects, which uses Bayesian estimation and a probabilistic model for the target support. Due to the underground targets being sparse, the B-scan is a sparse image. Using the CS principle, the stepped frequency radar is implemented using a subset of random frequencies at each antenna position. For image reconstruction we use Markov Chain and Markov Random Field models for the target support in the B-scan, where we also estimate the model parameters using the Expectation Maximization algorithm. The approach is tested using Web radar data obtained by measuring the signal responses scattered off land mine targets in a laboratory experimental setup. Our approach results in improved performance compared to the standard denoising algorithm for image reconstruction.
DOI10.1109/℡SIKS52058.2021.9606334
Citation Keykafedziski_compressive_2021