Interferometric SAR Data Processing Based On Wavelet Transform

The accuracy of SAR interferometry is effected by SAR system, geophysical parameters, e.g. earth surface condition and instable atmosphere. Algorithms of InSAR data processing are also key factors, which related directly to the quality of DEM and differential interferometric products, so it is necessary to develop advanced data processing algorithms and evaluate the accuracy of SAR interferometry. In this paper, wavelet theory is applied to InSAR data processing to improve the quality of SAR interferometry.Wavelet theory began from 1980's and developed greatly in 1990's. Wavelet theory provides a new powerful tool for studies relying on the time-frequency signal analysis. The main advantage of wavelet transform remains in its ability to locally describe signal frequency content. Unlike the short term spectral analysis, the wavelet transform provides localized frequential and spatial information about the signal. Small-scale details corresponding to high-frequency signals are represented on a wavelet basis of small spatial support, whereas large-scale variations are projected on wavelets with large spatial support. Wavelet is suitable to non-stationary signals processing. We applied wavelet transform to InSAR data processing and did following researches:1) Assumed the probability density function of signal and noise of real part and imaginary part of SAR interferogram.2)Tansformed the real part and imaginary part of interferogram to stationary wavelet domain, analyzed the PDF of signal and noise wavelet coefficients using Pearson distribution system, then evaluated the wavelet coefficients of signal in stationary wavelet domain by Maximum A Posterior(MAP) criteria to reconstruct the interferogram.3)Unwrapped interferogram after fringe edge detection using dyadic wavelet and edge linking by Active Contour model(Snake model).4)Analyzed the effects of atmospheric factors to InSAR DEM extraction and theturbulent character of atmospheric term.5)Estimated error power of noise and atmospheric term using fractal signal parameters estimation method based on wavelet", and then weighted combined multi-temporal InSAR data to eliminate atmospheric effects to DEM extraction. The results and conclusions are as following:1 )Based on Gaussian hypothesis for the signal and noise of real part and imaginary part of interferogram, in stationary wavelet domain, the PDF of wavelet coefficients of signal and noise term are analyzed to be PearsonlV type and Pearson VD type respectively.2)Set threshold in stationary wavelet domain to separate edge and noise, estimated wavelet coefficients in texture area to reconstruct the interferogram. The filtered results to a real interferogram showed the proposed method is powerful to interferogram speckle noise reduction, as well as it can preserve fine details in the interferogram that are directly related to the ground topography and maintain phase values distribution.3)For weak noise interferogram or filtered interferogram, dyadic wavelet can be used to detect fringe edges. A GVF snake model was used to link discrete edges, the linked edges were kept well accordance with fringes of phase map.4)Phase unwrapping algorithm based on edge detection is suitable to well separated fringes, the unwrapped phase can be rewrapped to the original phase map. which shows the result is correct.5)The atmospheric noise of interferogram is a fractal process, whose spatial spectrum exhibits the characteristic -8/3 powder law behavior associated with turbulence phenomena. The effects of this 1 / / process to InSAR DEM were anylysised and the error power of noise and atmospheric term were estimated for weighted combination of multi-temporal InSAR data. The test results showed this data fusion process can improve the accuracy of InSAR DEM.