Orthogonal Functions and Transforms
The Karhunen-Loève Transform
Also known as the Principal Component, Hotelling transform, or the Eigenvector transform (Ref: Hall).
This transform is based on statistical properties of the given image, which is treated as a random vector X.
where ?ij = E{(xi - ?i)(xj - ?j)}; ?j = E{xi}; i,j = 1, 2,…, n.