The Research On Multi-channel SAR Automatic Detection Of Slow Ground Moving Targets Based On The Interferogram

Aiming at the requirement of developing SAR-GMTI system, and in order to detect slow ground moving targets rapidly and efficiently from SAR images, this thesis deeply investigates the automatic multi-channel SAR techniques for slow ground moving target detection based on the interferogram. For the purpose of constructing practical techniques of automatically detecting slow ground moving targets, some key techniques such as designing the slow moving target metrics, statistically modeling the interferogram and metrics, automatically detecting slow moving targets by the interferogram are systematically examined according to theoretic analysis and experimental validation with the simulated data and real multi-channel SAR data.1) Deep analysis and development of the slow moving target metrics constructed by the decomposition and elements'transform of covariance matrix. By analyzing the basic principle of slow ground moving target detection upon multi-channel complex SAR images, summarizing present detecting methods, quantitatively studying and proving the effect of multi-look processing towards detection metrics, the thesis proposes the effective condition of the second-eigenvalue metric, modifies the similarity metric, and constructs effective eigen-projection metric and several new DPCA metrics.2) The statistics of the interferogram. Under the frame of multiplicative model, based on the complex Wishart distribution, and bringing in the theory of terrain classification brought forward by Frery et al., the new joint PDFs of interferogram's magnitude and phase (including I0_MP distribution, I_MP distribution and I1_MP distribution), interferometric real part and imaginary part (including R_I0 distribution, R_I distribution and R_I1 distribution), marginal PDFs of interferometric magnitude (including E-Gamma distribution, E-G0 distribution and E-G distribution) are deduced according to utilizing the reciprocal of a square root of Gamma and the square root of generalized inverse Gaussian law. Furthermore, the accurate and quick parameter estimators are constructed based on Mellin transform, containing E-Gamma_MoLC, E-G0_MoLC, E-G_MoLC and E2-G0_MoLC. The new PDFs and corresponding estimators make up the relatively complete statistics of the interferogram.3) A new automatic detector for slow ground moving targets based on dispersing characteristics of the interferogram. By the joint PDFs of interferometric magnitude and phase, real part and imaginary part, the new method can be implemented on the interferogram's magnitude-phase complex plane and real-imaginary complex plane. The new method can detect slow ground moving targets with high detection probability, low false alarm probability, the wider range of detection velocity, high stabilization and anti-jamming. Within the interferogram, this method can achieve an optimal detection performance.4) A new CFAR detector for slow ground moving targets based on marginal PDFs of interferometric magnitude and phase. By the proposed PDFs of interferometric magnitude and their corresponding accurate, quick parameter estimators, a new CFAR detector and its efficient implementation are designed. The detector has the characteristics of high practicability and excellent detection performance.5) A completely novel scheme for detecting slow ground moving targets based on the interferogram. By employing the back-scattering differences between slow ground moving targets and stationary terrain presented in the interferogram, a novel detection metric, named by M&P metric, is formed. And then, the statistics of M&P metric are deduced, containing S0 distribution, S distribution and S1 distribution. The effective parameter estimators of the three distributions, called by S0_MoLC, S_MoLC and S1_MoLC, are also built. For the implementation of the novel scheme, firstly, the different analytical expressions of the statistical distributions for acquiring CFAR thresholds are obtained. Secondly, the multi-look processing and the filter of M&P metric's global magnitude are introduced to improve the precision of the estimators. Thirdly, the new CFAR detector is combined. In a word, the novel scheme can universally and intelligently realize practical slow ground moving target detection.