PURPOSE: To recognize characters rapidly and punctuation marks, sonant marks, etc. precisely by reading out plural characters simultaneously, forming a square from the distribution of the dot data and defining the position in the square to identify these characters.

CONSTITUTION: A coordinate correcting part 2 checks the distribution 6 of lateral dot data within a scanned range and the distribution 7 in the longitudinal dot data of data inputted from an input part 1 and finds out a square surrounding the center character. Since the length of one side of the square is found out by said operation, the square corresponding to all the plural characters is formed. The square formed by said procedure is defined and corrected while referring information related to the row and line intervals of the printed character string previously registered in a recognition reference part 5. A comparator 3 compares the character data with the character dot pattern previously registered in the recognition reference part 5 and outputs the recognized character data to an output part 4.