MNF – Minimum Noise Fraction

Minimum Noise Fraction – MNF – looks for differences in variance in the multispectral data, similar to applying a PCA transform. The difference is that MNF whitens the data before applying the Principal Components Analysis transform.

MNF has three inputs, all of which are the same multiband image, but of different regions. The principal components are calculated from the region of interest specified in the “roi” image. The whitening is done based on the data from the “noise” image. The resulting PCA transform is applied to the “input” image. The three images must have the same number of bands, but they can be different sizes. Two cautions in using MNF. The larger the number of spectral bands that are included in the input images, the longer the computation time will be, and the output results can vary greatly as the “roi” and “noise” regions are varied.

The number of output bands is equal to the number of input bands. The output bands are ordered in decreasing variance, but since the input data is whitened before the transform is calculated, the output bands tend to vary in decreasing image quality.

In the “bandsOut” window, one can specify the number of output bands desired, which can be less than the number of input bands. If M output bands are specified, then bands 0 to M-1 will be output.