Speckle reduction of ultrasound images with anisotropic diffusion based on homogeneous region automatic selection

An adaptive selection method of diffusion threshold was proposed to improve the effectiveness and stability of a filter in speckle reduction of ultrasound images. An optimal threshold of the ultrasound image was determined by the Otsu binarization algorithm. Then, the ultrasound image was divided into blocks by Quad tree decomposition using the optimal threshold as the criterion of homogeneity. In descending order of the size, the present maximal blocks were picked up from the Quad tree decomposition result, and an optimal homogeneous region of the ultrasound image was selected by the proposed selection criteria. Finally, the diffusion threshold was obtained by analyzing statistical features of the optimal homogeneous region, and the ultrasound image was filtered using this diffusion threshold. The results demonstrate that the proposed method has better performance comparing with the Speckle Reducing Anisotropic Diffusion (SRAD) method and the Detail Preserving Anisotropic Diffusion (DPAD) method. It reduces the operation time effectively, and the average figure-of-merit by using the proposed method is 0.029, 0.129 higher than those by using other two mentioned methods. The proposed method avoids the manual selection of homogeneous area and can estimate the diffusion threshold accurately, which can reduce the speckles effectively while preserving the edges.