A matrix approach to character recognition

Abstract The problem of separating two relatively compact classes of matrices by a linear function is investigated. The modality properties of the corresponding optimality criteria are considered. It is recommended that this sort of classifier is used for separating matrix descriptions of characters. The classifier is easy to implement and the sample size is smaller than in the case of a common vector classifier. Some simulation results show that it is possible to utilize the theory of matrix classification in solving a special character recognition problem.