Based on the scale space correlation in wavelet domain, a new adaptive filtering algorithm is proposed to remove speckle in SAR images in this paper. This method decomposes SAR images into several scales with Stationary wavelet transform (SWT) and in every detail sub-image we use the idea of the scale space correlation to differentiate the edge information of detail from the noisy images. To deal with the edge information of points, the normal Wiener filter is used; to process the noisy points in homogeneous areas, the Bayesian soft threshold is applied to reduce speckle. After processed with this algorithm, SAR images can not only achieve more satisfied despeckling effect than other conventional methods but also show a good performance of preserving details and texture information. Additionally this algorithm can be more easily performed and has a lower computation burden so it has a widely applicability.
展开▼
