An anisotropic diffusion filter for reducing speckle noise of ultrasound images based on separability

Abstract

Anisotropic diffusion is being widely used in reducing speckle noise of ultrasound images. However, the traditional anisotropic diffusion algorithms are poor at preserving edges and usually make the image edges blurred when denoising, which negatively affects the following image analysis. In this paper, we modify the standard speckle reducing anisotropic diffusion to increase its ability of detecting edges and suppress the smooth at edge by using the separability of images. We extract contours from the original images, denoised images by SRAD and the images denoised by our proposed method, respectively. We analyze and compare the accuracy of these three kinds of contours. The result shows the proposed method performs better in edge-preserve and gets better images of high quality than SRAD, which can contribute to get more accurate contours.