Basic Shape Classification of Buried Object Using Pattern Matching in Ultrawideband Radar Image

This paper proposes the pattern matching technique to the signal pattern of windowed B-scan data to classify the object cross-section shape. The technique applies the dynamic time wrapping to adaptively measure the similarity of the object under test and the reference object. As studied case, we consider a metallic object with three basic shapes: circle, square, and triangle. To check the generality and performance of the proposed technique, we conduct the simulations and experimental studies with an ultra-wideband impulse radar. According to the results, the proposed technique is feasible enough, and it is applicable over the wide possible environments.