Research On Key Processing Techniques For Deriving Accurate DEM From InSAR

There are some prominent merits to derive DEM (Digital Elevation Model) and to monitor deformation of earth's surface by InSAR (Interferometric Synthetic Aperture Radar, or Synthetic Aperture Radar Interferometry) technique, as include speediness, high accuracy, full time, full weather, and large area observations, etc. Although it has become one of the most powerful techniques for earth observation, the state of practicality of InSAR all over the world is not satisfying due to the influence of data quality and some key processing techniques not being settled, such as the filtering of interferograms, the unwrapping of phases, etc. Aiming at the practical applications of InSAR to derive DEM, the key processing techniques were studied, such as the matching of master and slave images, the filtering of interferograms, the unwrapping of phases, and the estimation of baseline, etc. The applied InSAR-DEM software was developed. And the technique of multi-baseline InSAR was roughly studied to improve the rate of utilized data and the precision of derived DEM.The main work and innovations of this thesis are list as follows:1. The principles of SAR imaging and geometrical imaging models were presented and analyzed deeply. The basic principles and the common processing flowcharts of InSAR to derive DEM and of D-InSAR (Differential Interferometric Synthetic Aperture Radar) to monitor deformation of earth's surface were discussed.2. The algorithms and flowcharts of image matching for airborne dual antenna interferometric data and for space-borne repeated pass interferometric data were studied. The strategy of one-dimensional image matching in azimuth for master and slave images of InSAR was proposed. It can speed up the matching obviously. So the important breakthrough in one-dimensional image matching for Interferometric data was realized. The phenomenon of frequency offsets between master and slave images in range was analyzed. And the pre-filtering algorithms for master and slave images were presented.3. The filtering algorithms both in space and frequency domain were studied. The algorithms in space include multi-looking filter and vector filter. The algorithms in frequency include the spectrum weighting filter and the main frequency extraction filter. The Adaptive Zero Intermediate Frequency Vector (AZIFV) filtering algorithm was proposed for the first time. And the filtering algorithm based on vector decomposing and wavelet transformation was designed. These algorithms can realize the denoising and filtering of interferograms with dense fringes and with fringes whose frequency changes greatly. They can reduce the difficulty in phase unwrapping. In order to evaluate the phase preserving precision, a novel strategy by RMS (Root Mean Square errors) between the previous interferometric phases and the filtered phases was proposed. Combining with quality map and residue number. It can measure the filtering effects quantitatively and objectively.4. The phase unwrapping methods in common use for interferograms were introduced, such as the path tracking method and the minimum norm one. The adaptive phase unwrapping algorithm by combining quality map region growing, direct phase integral and mobile surface fitting for phases was proposed. It can settle the difficulty in phase unwrapping for low coherent areas. It considers the precision, the integrity and the consistency in phase unwrapping. The RMS of phases between the wrapped interferogram and the rewrapped one was proposed to evaluate the quality of phase unwrapping. It can do measurable evaluation efficiently for phase unwrapping.5. The baseline estimation methods by phases of flat earth and by phases of double height control points were studied and generalize. A new baseline estimation algorithm for cross orbits was designed with multiple height control points. The RFM (Rational Function Model) was utilized in the conversion of phases to heights. And the RFM between the unwrapped phases and the ground points' heights was proposed to realize the direct and accurate conversion of unwrapped phases to ground points' heights.6. The technique of multi-baseline InSAR was roughly studied. A new practical processing flowchart for multi-baseline InSAR was presented to increase the rate of utilized interferometric SAR data and to improve the reliability and precision of derived heights.7. The practical InSAR-DEM software system was coded by integrating the achievements in matching of master image and slave image, filtering of interferograms, phase unwrapping, baseline estimation and DEM reconstruction. Multiple sorts of InSAR data were used in a great deal of experiments of DEM reconstruction, as include the SIR-C/X-SAR data, the ERS-1/2 data and the airborne dual antenna data, etc. The experimental results show that the software is full in functions, steady in capability and reliable in precision. It lays basis for broad applications of InSAR to derive DEM.