JPS57143679A - Automatic character reader - Google Patents

Abstract

PURPOSE:To read accurately individual characters, by checking the number of black picture elements included around a white picture element and making a character string into a patterned blurred picture in accordance with this number and recognizing the position, the length, etc. of the character string on a basis of this picture and reading individual characters in the character string on a basis of these recognition results. CONSTITUTION:A binary picture consisting of white and black picture elements on the surface of a business form or a document 1 is picked up by an image pickup part 2, and the picture is supplied to a blurred picture generating part 3. All of the surface of the picture is scanned by this picture generating part 3; and if a white picture element is detected, the number of black picture elements included around this white picture element is checked, and a processing for changing black elements is performed in accordance with this number to make the character string into a patterned blurred picture. This blurred picture is scanned by a character string recognizing part 4 to recognize the position, the length, etc. of the character string, and these recognition results are given to a read head driving part 5 of a read head driving part 9, and individual characters are read by a read head 6. Thus, individual characters in the character string are cut accurately by a character cutting part 7 and are read accurately by a character recognizing part 8.