Open AccessNoise‐robust HRRP target recognition method via sparse‐low‐rank representation
Author(s) -
Li Long,
Liu Zheng
Publication year - 2017
Publication title -
electronics letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.375
H-Index - 146
ISSN - 1350-911X
DOI - 10.1049/el.2017.2960
Subject(s) - discriminative model , pattern recognition (psychology) , artificial intelligence , hinge loss , sparse approximation , computer science , noise (video) , rank (graph theory) , representation (politics) , noise reduction , signal to noise ratio (imaging) , radar , mathematics , support vector machine , image (mathematics) , telecommunications , combinatorics , politics , political science , law
A novel target recognition method is proposed for high‐range resolution profiles (HRRPs) of radar targets under low signal‐to‐noise ratio (SNR) conditions. This method achieves good recognition performance for noisy HRRPs with discriminative sparse‐low‐rank representation. The framework of this method is constructed based on sparse representation and low‐rank representation, which are applied to extract the local and global characteristics of target HRRPs. To guarantee the noise‐robust and highly discriminative features of the HRRPs, dictionary learning is adopted. In the training stage, a discriminative dictionary is produced based on hinge loss theory to improve the recognition performance. Denoising dictionary optimisation is implemented for noise suppression during the testing stage. Experimental results on measured HRRP data demonstrate that the proposed method can recover the original HRRPs and significantly improve the recognition performance for HRRP test samples under relatively low SNR conditions.