Naive Bayes nearest neighbor classification of ground moving targets

This work addresses the problem of automatic target recognition (ATR) using micro-Doppler information obtained by a low-resolution ground surveillance radar. An improved Naive Bayes nearest neighbor approach denoted as O² NBNN that was recently introduced for image classification, is adapted here to the radar target recognition problem. The original O² NBNN is further modified here by using a K-local hyperplane distance nearest neighbor (HKNN) instead of the plain nearest neighbor (1-NN) method. The proposed classifier outperforms minimum divergence (MD) based approaches with Gaussian mixture model (GMM). Performance of the proposed modified O² NBNN classifier was analyzed using collected radar measurements for variety of signal-to-noise (SNR) levels and sizes of training data.