Artificial Neural Network based automatic classification of target shapes from GPR data

Identifying and classifying buried artefacts remains a major scientific and technological problem. Shape recognition will aid investigators in classifying buried items and narrowing down areas of interest. An artificial neural network (ANN) model for automated object shape classification using GPR data is developed in this paper. A synthetic dataset is developed using finite difference time domain (FDTD) simulation and utilised to train and evaluate the proposed ANN model. The model performs well in identifying three shapes of cylindrical, rectangular, and triangular objects.