2D ground penetrating radar data identification

In this study, a classification method is proposed for a limited amount of ground penetrating radar data, based on a non-parametric approximation to Bayes decision probability. Ground penetrating radar is an electromagnetic sensor that detects objects both on the surface and under the surface. We use these preprocessed data for classification. The proposed study includes data enhancement, two different data reduction method that are PCA and time domain data reduction, a cepstrum analysis to extract feature and decision by k-nearest neighbor (k-NN).