Cluster driven anisotropic diffusion for speckle reduction in ultrasound images

In this paper, we propose a cluster-driven anisotropic diffusion (CDAD) filter for speckle reduction in ultrasound images. The proposed filter is based on the multiplicative noise model and is driven by K-means clustering algorithm. Instead of choosing homogeneous sample region with manual selection, the proposed algorithm is able to do it automatically (based on the clustering results). In addition, clustering result is used as a global characteristics descriptor to further improve the performance of noise removal as well as edge enhancement. The proposed filter was implemented and evaluated with real ultrasound images. Experimental results show that the proposed CDAD filter shows improved performance compared with Speckle Reducing Anisotropic Diffusion (SRAD).