This Master thesis introduces at first the basic principles of InSAR processing, and ends with the study of some of the most advanced topics in the InSAR domain to date, that is, PS technique and cross-interferometry. The main work and contributions of this Master thesis can be summarized as the follows: 1. Implementation of Azimuth Filtering: After a general description of Insar processing principles, we focus on the analysis on the Doppler Shift Phenomenon (Chapter 2). We point out that the difference of Doppler Centroid results in a decorrelation effect, and that this problem is to be solved by filtering processing. We have implemented the Azimuth Filtering algorithm and demonstrate on real images the improvement of the correlation after filtering. 2. PS selection and algorithmic techniques: The Chapter 3 first points out the lack and insufficiencies of existing PS selection methods; we thus propose a new adaptive threshold approach, able to cope with any dataset. In a second part of the Chapter, we introduce the commonly accepted analytical model of the phase measurement to retrieve the LOS (Line of Sight) velocity. We show that the system has multiple solutions; to overcome this problem, we developed an original optimization framework. The new PS technique as proposed in this Chapter has been applied to a set of ERS images covering Beijing area between 1995 and 1999. The subsidence rate has been estimated but could not be validated due to a lack of ground truth data. However, a comparison is made between the LOS velocity results obtained from different PS selection algorithms. 3. ERS-ENVISAT cross-interferometry: The theoretical basis of ERS-ENVISAT images cross-interferometry are presented in Chapter 4. Two difficulties raised by cross-interferometry are exposed in details: i) due to the difference of the signal wave lengths, the images pixel resolution is not identical in ERS and ENVISAT images ; ii) the phase difference equation, at the core of the interferometry technique, is augmented with an additional phase shift term, which needs to be removed before retrieving a “normal” interferogram. The first problem can introduce complication at the stage of image co-registration that generates the interferogram; to cope with this, we propose an efficient co-registration scheme, which is a generalization of existing methods. At last, we establish the relationship between the PS technique and the cross-interferometry, giving a brief introduction on PS continuity in ERS and ENVISAT image series.
修改评论