PURPOSE: To extract a correct significant character string out of a document where the horizontally and vertically written word strings are mixed together and 1 then to output this extracted character string to each of areas separated from each other via the ruled lines.
CONSTITUTION: An input document is divided into the areas separated from each other via the ruled line codes by a character extracting part 2. Then the code of each area is extracted and the vertically and horizontally written character strings including no blank are produced from those extracted codes. Then each character string is divided into words by a morpheme analyzing part 3 by reference to a word dictionary 1. Based on this dividing result, a deciding part 4 decides a character string including the smaller number of unknown words as a correct character string and outputs this string.