Hyperspectral unmixing is an image processing and data analysis technique used to decompose the mixed spectral measurements of pixels into their constituent pure material signatures, known as endmembers, and their corresponding fractional proportions, known as abundances. Because hyperspectral sensors capture electromagnetic radiation across hundreds of contiguous narrow wavelength bands, a single pixel often covers an area containing multiple distinct physical materials. To resolve this subpixel mixing, unmixing algorithms employ linear or nonlinear mixture models alongside computational methods such as non-negative matrix factorization, convex geometry, sparse regression, and deep learning. The resulting abundance maps and extracted spectral signatures enable precise material identification and quantification in fields such as satellite remote sensing, environmental monitoring, agriculture, and mineralogy.
