3 STAGE ML CLASSIFIER

A fast version of the maximum likelihood classifier is proposed and implemented in three stages. In each stage some groups are eliminated from the search process with the help of linear algebraic rules. The lower triangular matrix approach is used for the calculation of the quadratic term which requires fewer computations compared with the direct method. A theorem which defines the range of the quadratic term is used as a prime group elimination criterion in the group search process. Comparisons were made for computational requirements for one set of remotely sensed data. The proposed algorithm is observed to be three times as fast as the literal method.