Target classification based on micro-Doppler signatures

Abstract

In this paper, we propose a Gabor filtering method to extract localized micro-Doppler signatures represented in the time-frequency domain. The dimensionality of the extracted Gabor features is further reduced by using the principal component analysis (PCA) method. Therefore, a suitable classifier can be used for target classification based on their different motion dynamics. In our study, we use simulated radar data. Three different classifiers (Bayes linear, k-nearest neighbor, and support vector machine) are compared and tested. Our experiments show that Gabor features are robust in discriminating micro-Doppler effects of different types of micro-motions, and SVM classifier provides the best performance.