A New Statistical Model for Radar HRRP Target Recognition
6th International Symposium on Neural Networks, 2009 ~ 2009, pp 401-409, 2009
To resolve the problem of how to determine the proper number of the mixture models for radar high-resolution range profile (HRRP) target recognition. This paper develops a variational Bayesian mixture of factor analyzers (VBMFA) model. This method can automatically determine the optimal number of models by birth-death moves and can accurately describe the statistical characteristics of HRRP. So the VBMFA method should have better recognition performance than factor analysis and mixtures of factor analyzers method, and experimental results for measured data proved this conclusion.
Radar automatic target recognition (RATR); High-resolution range profile (HRRP); Variational Bayesian mixtures of factor analyzers (VBMFA); Variational Bayesian(VB); Mixtures of factor analyzers (MFA)