Median Modified Wiener Filter (MMWF) and MMWF* for 2D and 3D signal denoising
The Median Modified Wiener Filter (MMWF) was invented as an informational filter by Carlo Vittorio Cannistraci, and was proposed for the first time for application in denoising of two-dimensional gel electrophoresis maps in proteomics. The reference article is:
CV Cannistraci, FM Montevecchi, M Alessio
Proteomics 9 (21), 4908-4919
In a new article under review, the 3D extension of the median-modified-Wiener-filter (MMWF) and also its novel variation named MMWF* is proposed for the first time.
The performance of the new MMWF on 2D/3D non-stationary signals is even better, at least on our test datasets, than MMWF and wavelet-denoising. Noticeably, MMWF* gains stable high performance almost invariant to diverse window-size settings, which might represent a consistent advantage in automatic computational pipelines for denoising of non-stationary signals.
Below you can find the Matlab code for:
- the two-dimensional Median Modified Wiener Filter (MMWF_2D_website.m) proposed in the first published article
- the three-dimensional Median Modified Wiener Filter (MMWF_3D_website.m) proposed in a new article under review
- the novel two-dimensional variation of the MMWF that is named MMWF* (MMWF_2D_website.m) proposed in a new article under review
- the novel three-dimensional variation of the MMWF that is named MMWF* (MMWF_3D_website.m) proposed in a new article under review