Speckle noise reduction of ultrasound images using M-hand wavelet transform and Wiener filter in a homomorphic framework

In this paper we make use of M-band discrete wavelet transform (MDWT) and Wiener filter for speckle noise suppression of ultrasound images. Assuming multiplicative model for speckle noise, by applying logarithmic transform to noisy observation we convert multiplicative noise to additive one. Then, speckle noise can be reduced by taking the M-band wavelet transform of log-transformed observation and employing an adaptive thresholding technique. In order to improve the performance of this denosing approach, we propose a preprocessing stage which exploits Wiener filtering. Through this stage log-transformed observation is separated into two images and then each of these two images is denoised individually in M-band wavelet transform domain. The summation of these two images constructs the despeckled image. We have tested our method on realistic ultrasound images and compared the results with results of other speckle suppression methods.