A Density Current Modeled Adaptive Weighted Average Despeckling Filter for Ultrasound Thyroid Images

Speckle noise reduction is an important preprocessing stage for ultrasound medical image processing. In this paper, an adaptive weighted average despeckling algorithm is proposed based on density currents equation. An analogy between the gravity current density model for sea breeze and speckle noise affected image is used and a denoising filter is designed. Suitable threshold values to distinguish between different regions in the image are selected for the denoising filter by considering the structural similarity and mean square error parameters of the particular image. Hence the proposed denoising filter adaptively changes the threshold values depending on nature of image and magnitude of noise in the image. For each pixel, weights have been changed adaptively by considering local neighbours in different directions. Experimental results of proposed method for natural images, Field II simulated images and real ultrasound images, show that the proposed method outperforms other existing methods in terms of signal to noise ratio, Structural similarity index and Ultrasound despeckling assessment index. It is also shown that proposed filter is able to improve the visual quality of Ultrasound images and better preserves image structural details compared to existing methods.