Multivariate Change Detection in Multispectral, Multitemporal Images

This paper introduces a new orthogonal transfonn the multivariate change detection (MeD) transfonn based on an established multivariate statistical technique canonical correlation analysis. The theory for canonical correlation analysis is sketched and modified to be more directly applicable in our context. As opposed to traditional univariate change detection schemes our scheme transforms two sets of multivariate observations (e.g. two multispectral satellite images aquired at different points in time) into a difference between two linear combinations of the original variables explaining maximal change (Le. the difference explaining maximal variance) in all variables simultaneously. A case study using multispectral SPOT data from 1987 and 1989 covering coffee and pineapple plantations near Thika, Kiamhu District, Kenya, shows the usefulness of this new concept.