Reseach On Insar Imaging Algorithm Based On Bp And Multibaseline Phase Unwrapping

As an important research direction of synthetic aperture radar(SAR) technology,interferometric synthetic aperture radar(InSAR) has been widely used in three-dimensional terrain mapping, resource exploration and other fields. In this thesis, high-accuracy InSAR technology is researched. The specific work of this thesis are as follows:1、Study the InSAR registration method based on BP algorithm. Registration error is one of the major problems affecting interferometric phase accuracy during InSAR data processing. In this thesis, BP algorithm is applied to InSAR data processing. The registration algorithm based on BP is studied. By analyzing the projection coordinates in the master and slave images for the same scattering point,we can find BP algorithm can achieve coarse registration of the master and slave images。To further improve the accuracy of registration based on BP algorithm, this thesis have studied the influence of different imaging space on registration accuracy, we found that the registration accuracy will be higher, if imaging space is closer to the terrain in imaging area.2、Study the influence of BP on InSAR phase. Firstly, the thesis compares the difference between BP algorithm and the traditional imaging algorithm. By analzing the difference of the interferometric phase, we find that BP algorithm can remove the flat-earth-phase based on the registration accuracy. Since the BP algorithm can achieve coarse registration, so it has the ability to accurately remove the flat phase. Then, this thesis established the interferometric phase error model in the case of complex trajectory. The expression of the phase error is given. We find that the interferometric phase errors caused by attitude error is mainly in the flat-earth-phase. Since the IMU data can be used during BP imaging, it is possible to accurately compensate the motion error, and further improve the accuracy of flat-earth-phase removing. At the same time, we can get a better focus on the interferometric phase. Finally, the measured data is used to verify that BP algorithm can obtain high-precision interferometric phase.3、Research the multi-baseline phase unwrapping methods based on probability estimation. Firstly, the unwrapping method based on maximum likelihood(ML) estimation is studied. The ML method can unwrap the phase of complex terrain, but the accuracy is low. Secondly, in order to improve the accuracy of phase unwrapping, the multi-baseline unwrapping method based on the maximum posterior probability is studied. By introducing the relationship between neighboring pixels as a priori information, the maximum likelihood estimation problem is transformed into the maximum posterior probability problem. The graph cuts technology is also used to improve the accuracy of phase unwrapping. Then, we propose an improved algorithm for ML estimation phase unwrapping method. This method uses the peaks of the likelihood function and the relationships of adjacent pixel to optimize the result of ML estimation. The phase unwrapping accuracy of algorithm is high. Also it is a parallel algorithm, so the GPU can be used to improve the efficiency of phase unwrapping by parallel processing. The above unwrapping algorithms are validated by simulation data. Maximum posteriori probability estimation phase unwrapping algorithms and improved maximum likelihood estimation phase unwrapping algorithms have the similar unwrapping precision, which is much higher than the maximum likelihood estimation phase unwrapping algorithms. While the efficiency of improving maximum likelihood estimation phase unwrapping algorithms is better than the maximum posteriori probability estimation phase unwrapping algorithms.