Computational methods for processing ground penetrating radar data

摘要: The aim of this work was to investigate signal processing and analysis techniques for Ground Penetrating Radar (GPR) its use in civil engineering construction industry. GPR is the general term applied which employ radio waves, typically Mega Hertz Giga range, map structures features buried ground or manmade structures. measurements can suffer from large amount noise. This primarily caused by interference other radio-wave-emitting devices (e.g., cell phones, radios, etc.) that are present surrounding area system during data collection. In addition noise, presence clutter – reflections non-target objects underground vicinity target make measurement difficult understand interpret, even skilled human, analysts. This thesis concerned with improvements processes be order enhance detection characterisation process particularly multivariate techniques. Those include Principal Component Analysis (PCA) Independent (ICA). Both have been investigated, implemented compared regarding their abilities separate originating signals noise type data. Combination PCA ICA (SVDPICA) two-dimensional (2DPCA) specific approaches adopted further developed work. Ability those methods reduce unwanted investigated reported thesis, suggesting automated images a possibility. Further carried out concentrated on analysing performance at same time investigating possibility identifying characterising interest pre-processed images. driving idea behind part extract resonant modes individual traces each image properties poles characterise target. Three related but different Extended Prony, Linear Prediction Singular Value Decomposition Matrix Pencil methods. these approaches, technique has used dimensionality extracted compare measured various experimental setups. Performance shows offers best results.