A Novel Statistical Recognition Method Based on Hypersphere Model for Radar HRRP ATR
Lan Du, Hongwei Liu, Zheng Bao, Feng Chen · 2007
Different from general Gaussian-distributed data, L2normalized samples are applied to HRRP-based recognition to deal with the amplitude-scale sensitivity problem, therefore, geometrically speaking, power transformed HRRP samples spread on a unit hypersphere. This paper proposes a novel statistical recognition method for power transformed HRRP samples under the jointly multivariate Gaussian distribution hypothesis, in which the hyperspherical spread of HRRP samples and the effectively discriminating information contained in the noise subspace can be fully utilized without increasing computation complexity. The experimental results based on measured data show that our proposed method can greatly improve the recognition performance.