Feature Extraction in Radar Target Classification

Abstract

This paper presents experimental results of extracting features in the Radar Target Classification process using the J frequency band pulse radar. The feature extraction is based on frequency analysis methods, the discrete-time Fourier Transform (DFT) and Multiple Signal Characterisation (MUSIC), based on the detection of Doppler effect. The analysis has turned to the preference of DFT with implemented Hanning windowing function. We assumed to classify targets-vehicles into two classes, the wheeled vehicle and tracked vehicle. The results show that it is possible to classify them only while moving. The feature of the class results from a movement of moving parts of the vehicle. However, we have not found any feature to classify the wheeled and tracked vehicles while non-moving, although their engines are on.