An adaptive window hybrid median filter for despeckling of medical ultrasound images

Speckle is a random multiplicative noise which obscures the perception and extraction of fine details in ultrasound image and despeckling is necessary for better diagnosis. The main aim of despeckling is to keep the important features such as edges of the ultrasound image. In this paper an adaptive window hybrid median filter is proposed for despeckling of medical ultrasound images. In the proposed algorithm the size of the window of the hybrid median filter is selected based on the image region and to differentiate between the smooth and edge regions it uses Sobel edge operator. A window size of 3x3 is used if the center pixel to be processed is identified as an edge pixel and in smooth regions a window size of 5x5 is used. The proposed algorithm is tested on ultrasound image of the liver and synthetically speckled image and the performance is evaluated using image quality assessment parameters Peak Signal to Noise Ratio (PSNR) and Edge Preservation Factor (EPF). The results demonstrate that the proposed algorithm reduces the speckle noise effectively and also preserves edges.