In this thesis, a system for recognizing printed Korean characters from the scanned binary image which is based on the syntactic pattern recognition approach is described. Two pattern grammars are defined for accomplishing the recognition task: one for the alphabet symbol recognition, and the other for the character recognition (or composition). Web-grammar is used for the former, and string grammar for the latter. For the syntax analysis of the web-grammar, a top-down parser is constructed. YACC on the UNIX operating system is used for the automatic generation of the parser for the string grammar. This system has been implemented in "C" programming language on the VAX-11/780 machine running under the 4.2BSD UNIX operating system.