Wavelet-based despeckling for SAR image combing HMT model with GMRF model
Guozhong Chen, Xingzhao Liu · 2008
Speckle reduction is a prerequisite for many synthetic aperture radar (SAR) image processing tasks. In recent years, the hidden Markov tree (HMT) in wavelet domain is widely used for speckle reduction. The HMT model captures the persistence property of wavelet coefficients, but lacks the clustering property of them within a scale, whereas the Gaussian Markov random field (GMRF) model can characterize the intrascale contextual dependence of wavelet coefficients. In this letter, we propose a new wavelet-based despeckling method for SAR image by properly fusing the HMT and GMRF modelling firstly. Moreover, for better details preservation, we’ll borrow a parameter named multiscale local variation coefficient and develop two thresholds to measure the scene heterogeneity. The final experimental results for the simulated speckled images and real SAR images show that the proposed method can get better performance in terms of the extent of the speckle suppression and the fine details preservation.