摘要 |
PURPOSE:To precisely segment a character by printing plural characters in a specific area ahead of a document to be recognized at every several lines and estimating a character interval and a line interval based on binary image in a range corresponding to the specific area. CONSTITUTION:An input sheet 2 to which the specific character is printed in the specific area 14 the document to be recognized is printed in areas following the specific one 14, is inputted to a character observation part 1. It scans the input sheet 2, converts the contents of the specific area and the document into binary images and stored them in an image memory. Based on a binary image in the range corresponding to the specific area 14 in an image memory 3, a line interval estimation part 4 and a character interval estimation part 5 estimate the line interval Y0 and the character interval X0, respectively. The character interval estimation part 5 obtains length from the left end of the specific character and the end of the next character, or length from the center of the character in the line direction to the center of the next character, averages the lengths and calculates the character interval X0. With the aid of the line interval Y0 and the character interval X0, characters are segmented.
|