| 摘要 |
PROBLEM TO BE SOLVED: To allow accurate character recognition by accurately estimating a character position in a character string image.SOLUTION: A first character area extraction unit 12 extracts, per character, character images representing characters from a character string image. A collation unit 16 collates the character images with character information stored in a database to output characters represented by character information of the highest matching degree, for each character image. A reliability determination unit 18 determines whether each of the output characters is reliable or not on the basis of an index indicative of a likelihood of the character and a size of the character. A character position estimation unit 20 designates, on the basis of the size of a character determined to be reliable, sizes and positions of the other characters. A second character area extraction unit 22 extracts a character image for each character on the basis of the estimated positions and sizes of the other characters. The collation unit 16 collates again the respective extracted character images with character information to output again a character represented by character information of the highest matching degree, for each character image. |
