Software

Maximum noise fraction (MNF) is a BSS-variant technique. This approach was originally developed to remove noise from multispectral satellite images and subsequently extended to remove noise from time-series. It assumes that an observed signal is being linearly mixed with a noise source which is uncorrelated with the true signal. Although MNF is similar to PCA in that both rely on second order moments there is an essential difference between the two methods: PCA derives the eigenvectors and eigenvalues of the covariance matrix of the observed signals under the constraint that it maximizes the variance explained by each of the components. In contrast, MNF derives generalized eigenvectors and eigenvalues for the signals and the noise covariance matrices.

The Qt-application is an app container. It can integrate different applications existed and developed by yourself or others.