Speckle Filtering of Ultrasound Images Using Improved Wavelet Shrinkage Guided Filter

Speckle noise is an intrinsic yet undesirable residual part of the medical ultrasound images, which considerably degrades the visual quality and limits the accuracy of automatic diagnostic techniques. Hence speckle suppression is an essential task before the examination and processing of the medical ultrasound images. Aiming at the problem of multiplicative speckle noise, an improved despeckled method is proposed in this paper, which is based on WaveletShrinkage_Guided filter. The proposed method used the Daubechies20 (d20) wavelet transformation to decompose the ultrasound images and an improved wavelet shrinkage algorithm to filter the high-frequency component. The wavelet coefficients of low-frequency sub-bands are filtered by an improved guided filter with varying regularization parameter and the inverse DWT is used to obtain the noise free image. Synthetic image and an ultrasound image experiments with the comparison of the Guided filter and WaveletShrinkage_Guided filter are carried out. The quantitative results of the proposed strategy show, that it outperforms the other despeckling methods in terms of favorable speckle suppression and edge preserving factors.