Radar Automatic Target Recognition Based on Complex High-Resolution Range Profiles

• Published in: Proceedings of 2006 CIE International Conference on Radar : Oct. 16-19, 2006, Shanghai, China : RADAR 2006 pp. 1 - 5
• Main Authors: Lan Du, Hongwei Liu, Zheng Bao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2006
• Subjects: Baseband; complex high-resolution range profile (complex HRRP); Data mining; Frequency; initial phase sensitivity; Light scattering; minimum reconstruction error approximation; Principal component analysis; principal component analysis (PCA); Radar automatic target recognition (RATR); Radar scattering; Radar signal processing; Signal processing algorithms; Target recognition; Testing; time-shift sensitivity
• ISBN: 0780395824; 9780780395824
• DOI: 10.1109/ICR.2006.343562

Radar high-resolution range profile (HRRP) has received intensive attention from the radar automatic target recognition (RATR) community. Usually, complex HRRPs are not fully used for RATR but only their amplitude vectors, while the phase information of them is discarded due to the fact that the initial phase of a complex HRRP is strongly sensitive to target position variation. However, the phase information of complex HRRPs may also contain valuable target discriminant information, which may further improve the recognition performance. This paper concerns RATR using complex HRRPs. To' deal with the initial phase sensitivity of complex HRRPs, we extract the complex HRRPs' feature subspace within each target-aspect sector of each target via principal component analysis (PCA) as the corresponding template during the training phase, while in the test phase, project the test sample onto each feature subspace and search the optimal approximation of the test sample with the minimum reconstruction error to decide which target the test sample belongs to. It is shown that the whole process is independent of the initial phases of complex HRRPs. Furthermore, to make the proposed recognition method more practical, a fast time-shift compensation algorithm is proposed. In the recognition experiments based on measured data, the proposed recognition method using complex HRRPs achieves better recognition results than that using only the amplitude vectors of the complex HRRPs