Noise Reduction of the Radar Interferometric Phase Used for Surface Displacement Measurement Based on Wavelet Transform in the Complex Domain

Interferometric Synthetic Aperture Radar (InSAR) technique using phase information has demonstrated its abilities in topographic mapping and measuring surface deformation with the precision of meter and sub-centimeter, respectively in a very high spatial resolution. However, various limiting factors such as spatial and temporal decorrelation, atmospheric effects and thermal noise of the radar sensor introduce different types of noise into the interferograms, which makes the phase unwrapping too difficult to obtain the accurate results. In this study, an algorithm for noise suppression is presented based on wavelet transform in the complex domain. The high-frequency data due to the phase jumps is not appeared in the complex domain. Therefore, the wavelet coefficients obtained in the complex domain include mostly the noise. The wavelet coefficients of the noisy interferogram are then filtered using the threshold computed from the related wavelet band. In comparison with the other noise reduction methods such as multi-look processing and those based on Fourier transform in small windows, the proposed algorithm can reduce the noise while keeping the spatial resolution without the need for windowing the interferogram. Quantitative and qualitative evaluations of the results obtained by the new method applied on the simulated and real noisy data show high performance of the wavelet transform in noise reduction.