Study On Methods For SAR-GMTI

Synthetic aperture radar and ground moving target indication (SAR-GMTI) combines the capability of high resolution earth observation and moving target detection and relocation, which is critical to military and civil application. High resolution imaging of stationary targets is achieved depending on the motion of platform, while the stationary targets are just regarded as clutter for GMTI mission, and clutter suppression is a prerequisite for moving target indication. The raw echoes cannot be coherently integrated due to two factors. On one hand, SAR transmits pulses repeatedly during the motion, which can be treated as discrete sampling in slow time and causes Doppler ambiguity. On the other hand, the fast moving targets will cause large range cell migration. Thus the detection performance of fast moving targets is limited. This thesis exploits a few kinds of strategies to deal with the problems resulting from real data processing, such as multi-channel SAR-GMTI clutter suppression and fast moving target detection as well as imaging. The outline of the thesis is listed as follows:1. The coherence enhancement is the prerequisite of multi-channel SAR moving target detection under strong clutter condition. A channel balancing technique based on the two-stage filters in different raw data domain is presented to overcome the differences in amplitude and phase. The amplitude and phase filtering in range frequency and azimuth time domain is used for range frequency response difference, while the amplitude-phase filtering in range-Doppler domain for Doppler spectrum difference. The first step of filtering removes the deterministic difference by using an adaptive filter in district with high coherence, and a non-adaptive one in the noisy district. The second step of filtering is used to suppress the noise further. After the channel balancing the clutter can be well cancelled and the moving target is obtained. The effectiveness of the channel balancing algorithm carried out in the raw data domain is demonstrated with the measured airborne data.2. In order to break through the maximum detectable velocity limitation due to the limited SAR pulse repetition frequency (PRF), two fast moving target indication methods are developed. One method utilizes the relation that the line slope of target trajectory is proportional to the ambiguity number of fast moving target in the range frequency and azimuth compressed time (RFAC) domain, and the ambiguity number can be obtained from Radon transform. The other method exploits the relation between range envelope and the ambiguity number in range compressed and azimuth time domain, following with the Keystone transform. Both the approaches acquire moving target imaging via 2-D matched filtering. The results of simulation and real data processing show the efficiency of the approaches.3. SAR imaging of the detected moving target is fundamental for target identification and very significant for military application. The radial velocity of fast moving target will cause the migration through range cell as well as Doppler ambiguity in azimuth, and the along-track velocity will cause the variation of the Doppler chirp rate. Both factors will make the echoes hardly be integrated coherently and the point spread function (PSF) of moving target in SAR image will become defocused. An approach based on Keystone-Wigner transform is proposed to implement the fast moving target imaging. Firstly, the large range cell migration is corrected in range compressed and azimuth time domain, detection is carried out after the coarse focusing in azimuth. Then the moving target signal is transformed back into RCAT domain, where the KWT is performed. Lastly, the image of fast moving target and the Doppler parameters are obtained. The motion parameters are estimated simultaneously with the fast moving target refocused. The proposed method can robustly estimate the motion parameters of the fast moving targets, which are illustrated with both simulated and real data.4. The development of sparsity theory has attracted wide public concern in the field of signal processing in recent years. Due to the perfect performance of sparse signal representation, a ground moving target imaging method based on sparse signal representation is proposed. In radar echoes the moving target signals take up the minority compared with the background clutter scatterers, so after clutter rejection the moving target signals can be treated as sparse LFM signals, which can be reconstructed via sparse representation on the overcomplete basis. A super-resolution imaging can be achieved by this method. The simulation and real data experiments show the validity of the approach.5. The capability of SAR high-resolution imaging with wide swath is an important aspect for earth observation satellites. A ground/sea surface moving target indication approach is developed with single channel SAR system under low PRF condition. This approach takes both the Doppler centroid and Doppler spectrum ambiguities into account. It can increase the maximum detectable velocity without the limitation of the PRF by using the multi-look beat frequency technique to finish radial velocity estimation of the moving target. The time-frequency property is used to estimate the target location. Thus the mission of the parameter estimation and target relocation can be achieved. The effectiveness of the proposed method is validated by the simulation and real data processing.6. The main advantage of bistatic SAR over monostatic one is that it can obtain more rich scattering information, thus more available for target identification. Space-surface multi-channel SAR-GMTI system has the capabilities of both clutter suppression and viewing angle flexibility. A multi-channel SAR-GMTI processing method under the space-surface generalized bistatic configuration is proposed. The basic moving target indication equations are presented, and the coherence enhancement method between the SAR images derived from different channels and the clutter character are discussed. The translation variant is dealt with using a phase compensation method. Simulation result shows the feasibility of the proposed method.