Accession Number : AD0692157

Title :   ALGORITHMS FOR SELECTING DISCRIMINATION FRAGMENTS FOR A SET OF PRINTED CHARACTERS REPRESENTED BY STATISTICAL REFERENCE PATTERNS,

Corporate Author : FOREIGN TECHNOLOGY DIV WRIGHT-PATTERSON AFB OHIO

Personal Author(s) : Avrukh,M. L.

Report Date : 02 JUN 1969

Pagination or Media Count : 25

Abstract : Two versions of an algorithm are described for selecting the discrimination fragments for a set of standard symbols represented by their statistical reference patterns. These algorithms have certain advantages over other algorithms because the use of discrimination fragments increases the possibility of discriminating between character parameters which are very similar but which differ in fine details. The use of statistical reference patterns lowers the requirements in regard to the quality of print and the exact position of the matrix with respect to the unknown character. It also mades it possible to unify characters which are printed in different style and are of different size (by reducing them to a common scale).

Descriptors :   (*READING MACHINES, CHARACTER RECOGNITION), (*CHARACTER RECOGNITION, ALGORITHMS), STATISTICAL ANALYSIS, INPUT OUTPUT DEVICES, USSR

Subject Categories : Computer Hardware

Distribution Statement : APPROVED FOR PUBLIC RELEASE