Local Coherence based Fast Speckle Reducing Anisotropic Diffusion

1 Computer Science College, Sichuan University Abstract. In ultrasound image, because of the presence of speckle, the image contrast resolution is reduced. And the image speckle noise also limits the effective application of image processing and analysis methods such as edge detection and segmentation methods. In recent years, a class of new methods called anisotropic diffusion has been developed in the field of ultrasound image speckle reduction. From the aspect of the quality of ultrasound speckle reduction, this class of method is desirable, however, most of this kind of methods are slow in speed. In this paper, we propose a new anisotropic diffusion based speckle reduction method which uses local coherence to control diffusivity in each pixel. We use additive operator splitting method to discretize our method to reduce the time consumption, Comparing with traditional methods, our method is faster in speed and, at the same time, its smoothing result is acceptable in terms of image quality.