Fast reduction of speckle noise in real ultrasound images

In this paper, we concentrate on fast removing speckle noise in real ultrasound images. It is hard to design a fast algorithm to solve a speckle reduction model, since the data fitting term in a speckle reduction model is usually not convex. In this paper, we present a convex variational model to deal with speckle noise in real ultrasound images. The data-fitting term of the proposed model is obtained by using a generalized Kullback–Leibler distance. To fast solve the proposed model, we incorporate variable splitting method and Bregman iterative method to propose a fast ultrasound speckle reduction algorithm. The capability of the proposed method is shown both on synthetic images and real ultrasound images.

► A convex model is proposed to remove speckle noise in real ultrasound image. ► A generalized KL divergence is introduced as discrepancy measure between two images. ► Variable splitting and Bregman iterative method are incorporated to fast solve the model. ► Three non-convex and two convex despeckling models are compared to show the efficiency.