An algorithm to reduce clutter and estimate the dimensions as well as material of the landmine has been proposed. Finite Difference Time Domain (FDTD) algorithm is used in order to simulate ground penetrating radar (GPR) for landmine detection. The soil is modelled considering its inhomogeneous nature and various minute anomalies which contribute as false targets. Plastic and Metal landmines have been simulated and the proposed algorithm is successful in differentiating the two. All types of possible clutter, antenna crosstalk, air-ground bounce, false target, Additive White Gaussian Noise(AWGN) and reflections have been simulated. Statistical technique, Principle Component Analysis (PCA) is used to reduce AWGN and remove false targets. A novel method to remove reflections has been proposed and discussed in detail. A correlation between the number of reflections and the material of the mine has been seen and the same has been envisaged to estimate the material. Different scenarios and real world conditions have been simulated and the proposed algorithm has been successful in detecting and classifying the mines with negligible error.
rajathkmp/LandmineLocalizationGPR
An algorithm that performs localization of landmines and estimates the material of the same.
Jupyter Notebook