Spectrum Estimation Research With Application In Radar High Resolution Imaging

The radar imaging technology is very important in the development of modern radar technology and very valuable in military and civil application, especially in precision guidance application. The high resolution range profile(HRRP) and the SAR two-dimensional resolution images contain the information of objects' structure and distribution. Therefore, it owns sound prospect in object identification, object resolution and selection, and high accuracy tracking application.In radar precision guidance, the wave beam duration time is extremely limited and the data samples used for object imaging is less, therefore in this condition, in order to obtain object information quickly and accurately, it requires higher standards for radar imaging algorithm. The imaging methods based on the spectrum analysis of high order cumulant, has been widely utilized for its good noise-suppression capability. While in the condition of small sample collection, the traditional spectrum analysis method of high order cumulant have the problem of signal model distortion, which affects the effect of spectrum estimation. In order to solve this problem, this paper presents a novel high order statistics which can be used for harmonic spectrum estimation in small sample collection condition, and establishes the signal processing model based on this statistics for harmonic spectrum estimation. Experimental results show that this statistics has good effect in harmonic parameter estimation.For high-resolution radar imaging, the radial range resolution of DFT based high-resolution algorithm is restricted by the bandwidth of radar transmitting signal, and the cross range resolution is restricted by the beam duration time. While the pseudo-spectral function based super-resolution algorithm has good range resolution but it can not estimate true amplitude and phase of all the frequency components, which is a disadvantage for angle measurement in amplitude comparison or phase comparison mono-pulse tracking radar. This paper presents a cumulant based Max likelihood estimation method, which can estimate the amplitude and phase of multiple scattering centers distributed in resolution cells and retains good frequency spectrum resolution ability of super-resolution methods.Data extrapolation method has super-resolution ability in spectrum estimation, but the traditional DFT based data extrapolation method is time-consuming because of iterative calculation. This paper proposes a new data extrapolation based spectrum estimation method. In this method, the start and end frequency of the band limited signal spectrum is firstly achieved by conventional DFT analysis, then the frequency domain high sampling rate discrete spectrum is estimated by using Max likelihood method. The method has good performance in resolution, amplitude estimation, and avoids time-consuming iterative computation.