Using Heavy-Tailed Levy Model in Nonsubsam pled Shearlet Transform Domain for Ultrasound Image Despeckling

F or any coherent imaging systems including ultrasound, synthhetic aperture radar and optiical laser, the multiplicative speckle noise de grades bo th the spatial and contrast resolution of the image. So, speckle suppression or despecklinng is necessary before processing like image seggmentation, edge d etection, and in ge neral any medical diagnosis. It is quite a min d-numbing task to analyze the corru pted imag es. Amongg many methods th at have beeen proposed to perform this task either in spatial domain or in transformed domain, there exists a class of approaches that use coefficient moddelling in transform domain. The pur pose of the paper i s developiing a novel despeckling methhod in noonsubsampled shearlet tran sform (N SST) based on coefficient modelling. B ayesian maximum a posteriori (MAP ) estimator is used where heavy-tailed Lévy (HTL) disstribution is assumed for estimating the noise-free NSST coefficients. The main contribution of this paper is connsidering HTL for modeling the NSST coefficients for the first time becausee of its low computational complexity. The proposed algorith m maintains a balance bet ween speckle suppre ssion and feature preserva tion. Finally, exp eriments show that the propo sed met hod outperforms others in terms of visual evaluation and assessment par ameters.