Study On Signal Processing Technique Of Spaceborne InSAR

Spaceborne InSAR system is a novel spaceborne radar system combining satelites formation flying technology with SAR technology. It measures the ground elevation by repeat track interferometry of one satellite or several satelites flying and working in formation. Compared with other DEM measuring methods and other InSAR systems, Distributed-Spaceborne-InSAR System attracts more concern around the world for its special advantages. Aiming at this new thesis,this paper studied the signal processing methods of spaceborne InSAR, including complex images coregistration, shadow and layover detection, phase diagram filtering and unwrapping. Especially, concerning some complex conditions, this paper has proposed some several novel processing methods that provided with more universality. The work of this paper will contributes much for the implementation and application of distributed spaceborne InSAR system.Complex images coregistration was studied in chapter 2. The coregistration of spaceborne InSAR system can commonly divided into coarse coregistration and fine coregistration, and further the coarse coregistration includes pixel and sub-pixel coregistration. Concerning coregistration methods, coarse coregistration can be achieved by geometrical method, manual method and correlative method, and the fine coregistration can be done by methods based on correlation and maximum spectrum. While the coherence is related to many factors, such as SNR, system geometry, baseline, the terrain etc. So when the coherence of the images is very low and the deformation between the images is large and complex, coregistration can nolonger be done by the traditional methods. Thus, a new method based on corner points detection, wavelet-Pyramid and Delaunay Triangulation is proposed for this "noncoherence" images. The registration accuracy of the method is at 0.1 pixel level.Shadow and layover detecting method were studied in chapter 3. There are often some regions where no radar echo received or several echos mixed together in InSAR images, the former was named "shadow", the latter was named "layover". Both shadow and layover will destroy the coherence of InSAR phase diagram, and put obstacle to phase unwrapping. Based on the analysis of the characteristic of shadow and layover, the paper proposed a novel detecting method of InSAR shadow and layover, which makes use of local frequency estimation, segmentation of signal amplitude, and morphological processing. The shadow and layover region detected can be used for weight parameters setting and other processing steps after.Phase filtering and unwrapping method were studied in chapter 4. The paper analysed the noise sources and compared the performance of several filtering method through experiments. A overview on InSAR phase unwrapping method was made in this chaper, all methods were divided into 3 sorts: "path-following", "minimum-norm" and "network flow".When doing the local frequency estimation, we found a encouraging face that high-accuracy local frequency can replace the phase difference of phase diagram and be used in phase unwrapping, and it is much more accurate than the filtered phase difference. Noticing this, a novel phase unwrapping method based on self-adapting window local frequency estimation and multigrid technique was proposed in this chaper. Thanks to using the local frequency, the traditional phase filtering step is nolonger needed, and shadow and layover detection can be done the same time, the processing time can be reduced meanwhile. It was proved by experiments that the phase unwrapping method proposed in this paper can provide a more accurate result compared with other traditional methods.