Automatic target classifier for a Ground Surveillance Radar using linear discriminant analysis and Logistic regression

This paper presents the design of an automatic target classifier for a Ground Surveillance Radar namely NUST Radar* (NR-V3). The classifier is developed to distinguish between pedestrians, vehicles and no target (noise) classes. Feature vectors are extracted from the FFT spectrum of radar audio signal. Logistic regression and linear discriminant analysis based classifiers are used for classification of feature vectors. The classifiers are trained and tested using radar data collected with NR-V3. Overall classification accuracy of 95.6% and 92% is achieved for Logistic regression and linear discriminant analysis classifiers respectively.

[1]  Jaakko Astola,et al.  Ground moving target classification by using DCT coefficients extracted from micro-Doppler radar signatures and artificial neuron network , 2011, 2011 MICROWAVES, RADAR AND REMOTE SENSING SYMPOSIUM.

[2]  A. G. Stove,et al.  A Doppler-based target classifier using linear discriminants and principal components , 2003, 2003 Proceedings of the International Conference on Radar (IEEE Cat. No.03EX695).

[3]  Jonathon Shlens,et al.  A Tutorial on Principal Component Analysis , 2014, ArXiv.

[4]  Henry Leung,et al.  Classification of audio radar signals using radial basis function neural networks , 2003, IEEE Trans. Instrum. Meas..

[5]  M. Andric,et al.  Ground Surveillance Radar Target Classification Based On Fuzzy Logic Approach , 2005, EUROCON 2005 - The International Conference on "Computer as a Tool".

[6]  Keinosuke Fukunaga,et al.  Introduction to Statistical Pattern Recognition , 1972 .

[7]  A. Cohen,et al.  GMM-based target classification for ground surveillance Doppler radar , 2006, IEEE Transactions on Aerospace and Electronic Systems.