Identification and classification of reflected landmine signals based on complex resonance frequency in dispersive media

This paper presents a new feature for the reflected landmine signals at different depths. It can be described mathematically by applying Prony's method, to calculate the complex resonance frequencies (CNR), which are considered as suitable features to discriminate different targets. Different classification techniques were evaluated: artificial neural network (ANN), K-Nearest Neighbour (KNN), Multi-ClassSupport Vector Machine (MC-SVM) and Decision Tree(DT).