The number of spectral bands assocuates with a remote sensing system is referred to as its data dimensionality. Hyperspectral remote sensing systems such as AVIRIS ans MODIS obtain data in 224 and 36 bands, respectively. The greater the number of bands in a dataset (i.e., its dimensionality), the more pixels that must be stored and processed by the digital image processing system. Storage and processing consume valuable resources. It is necessary to reduce the dimensionality of hyperspectral data while retaining the information content inherent in the image.
A method for dimensionality reduction in hyperspectral data and minimizing the noise in the imagery is the minimum noise fraction (MNF) transformation. The purpose is to minimize the noise in the imagery, i.e. to identify noise and segregate it from true information, and to colaps the useful information into a much smaller set of MNF images. The MNF transformation applies two cascaded principal components analyses.