Maximum likelihood discriminant analysis on the plane using a Markovian model of spatial context

Abstract A method is presented for utilising spatial information in performing discriminant analysis on multivariate data at each point on a regular lattice, as for example with LANDSAT. The spatial information is modelled as a particularly simple Markov random field. The discriminant analysis is based on the maximum likelihood method. The results are compared with alternatives by Monte Carlo simulation, and seem to be encouraging.