Radar automatic targt recognition for alterable noise environment

Noise level of radar target returned echoes is an essential issue for HRRP automatic target recognition, which will deteriorate recognition performance if test sample have different noise level compared with the training samples. Assuming that HRRP contains additive Gaussian white noise and HRRP signals of range cells are jointly Gaussian-distributed, this paper selects PPCA-subspace model to describe HRRP. The main contribution is a promising adaptive method to deal with alterable noise environment between training phase and test phase for HRRP statistical recognition. To make the algorithm more practical, an approximate algorithm is presented to accelerate the original one while keeping the sacrifice of recognition precision very small. Simulated recognition experiments based on measured data illustrate our proposed method´s effectiveness.