Micro-Doppler based target classification using multi-feature integration

Three novel micro-Doppler feature extraction algorithms are presented and applied to a dataset containing real X-band radar data of moving ground targets. In each case data dimensional reduction was carried out using principal component analysis (PCA) and incorporated into the feature extraction process. Extracted features are classified using a support vector machine (SVM) classifier. It was found that all three algorithms were able to produce classification accuracies in excess of 90%. The performance of the different algorithms are shown to depend on the method used and the degree of dimensionality reduction imposed at the PCA stage.