Accession Number : AD0693595
Title : AN ADEQUACY THEOREM FOR AN OBJECT AND ITS RECOGNITION SYSTEM,
Corporate Author : FOREIGN TECHNOLOGY DIV WRIGHT-PATTERSON AFB OHIO
Personal Author(s) : Gergely,T.
Report Date : 14 JUL 1969
Pagination or Media Count : 15
Abstract : Application of the estimator theory (in mathematical statistics) to the selection of an optimal set of prototypes is considered; the optimum corresponds to minimum error and maximum resolution. The theorem reads: 'To obtain an optimal recognition, with a certain definite-criterion algorithm which is a measure of dispersion, it is necessary and sufficient to work out - during the recognition-system learning - an optimal set of prototypes, i.e., a set of such prototypes whose space is a most accurate model of the observation space and corresponds to the space of the external medium.' It is demonstrated that the optimal set of prototypes is adequate to the external medium. In recognizing a check pattern, the recognition system uses only an active part of the entire set of prototypes. The potency of this active subset depends on the recognition-algorithm criterion.
Descriptors : (*PATTERN RECOGNITION, STATISTICAL ANALYSIS), IMAGES, CLASSIFICATION, SAMPLING, ALGORITHMS, ERRORS, CONFIDENCE LIMITS, SET THEORY, AUTOMATION, CONTROL, PROBABILITY DENSITY FUNCTIONS, USSR
Subject Categories : Cybernetics
Distribution Statement : APPROVED FOR PUBLIC RELEASE