SID – Spectral Angle Map

Spectral Information Divergence – SID – measures the spectral divergence between each pixel of the input image and one or more image references of the same spectral dimension, i.e. the same number of spectral bands, by measuring the distance between the probability measures of the two spectra.

SID has two inputs, an input image and a reference image. Both images must have the same spectral dimension, e. the same number of spectral bands. The reference image may be cropped to a very small region, or it may be the whole image and the crop window is specified in SID’s arguments. If more than one window is specified in the “roi” (region of interest) window, then all of those regions are part of the reference. Under the “regions” option, one can choose to average all regions together to for one reference, or they can be used separately and each window produces an individual output.

If there is only one reference window in “roi”, then the output will have only one band. If multiple windows are specified and the choice is to use them separately, then the output will have one band per window.

The value of each output pixel varies with the similarity of the two spectra. A value closer to 0 represents a pixel that has a similar spectrum to the reference spectrum.