Novel speckle remove method based on ASUSAN nonlinear diffusion

As we all know, speckle noise exist in ultrasound and radar images, and it brings bad affection to image analysis, like image segmentation, image fusion and image registration. So far, many diffusion methods have been put forward to remove speckle noise, but they all have some defections because their diffusion models are unsuitable for speckle noise. In this paper, a novel speckle remove method based on ASUSAN nonlinear diffusion is proposed. We adopt SUSAN operator to detect image edges, then k-means algorithm is used to select fully formed speckle region, finally our ASUSAN diffusion method is obtained to remove speckle noise by iterations. The experiments demonstrate that our proposed method achieves better results in PSNR, MES, FOM and image edges preserve, compared with the other methods.