Short-Time Fourier Transform and Wigner-Ville Transform for Ultrasound Image De-Noising through Dynamic Mask Thresholding

摘要

New approaches to filter out multiplicative noise from ultrasound medical images are presented in this paper. A performance comparison is made between the Short Time Fourier Transform (STFT) and the Wigner-Ville Transform (WVT). After segmenting the image into small, overlapping, dyadic lengths segments, STFT or WVT is applied to each segment individually. A minimum number of coefficients per STFT or WVT time-frequency plane is used, which has been found sufficient to represent the entire time-frequency plane. When applied on simulated and real ultrasound images, these approaches have outperformed popular nonlinear de-noising techniques, such as Wavelets, Total Variation Filtering and Anisotropic Diffusion Filtering. STFT provided the maximum cleaning of speckle noise, while WVT preserved the image edges and provided maximum resolution.