Discrete Shearlet Transform Based Speckle Noise Removal in Ultrasound Images

Abstract

Speckle noise removal plays an important role in ultrasound image diagnosis. Existing speckle noise removal techniques have drawbacks such as lose of edge information, texture information and inability to remove low frequency noise. To overcome these issues, a new discrete shearlet transform (DST) with nonlinear diffusion is proposed in this paper. DST comprises of localization, directionality and multiscale features which are crucial for ultrasound despeckling. Adaptive thresholding and nonlinear diffusion are combined with DST to remove both high frequency and low frequency noises so that the superior filtering performance can be achieved. The comparison performance of the proposed method with other despeckling techniques indicates the superiority of this technique.