| 摘要 |
<P>PROBLEM TO BE SOLVED: To enable a difference between printed characters in a plurality of document images to be easily grasped. <P>SOLUTION: Difference regions of parts having a difference in two images to be compared are extracted by a difference region extraction unit 101. Characters of two character images in character regions having a difference are respectively recognized by a character recognition unit 102. Specifically, it is determined which character image is close to the input character images with reference to data in a character pattern database 105 to obtain corresponding character codes. Next, the character codes are compared by a character comparison unit 103, and when the character codes are the same, a result is output as a font difference. When the character codes are not the same, a result is output as a difference in characters themselves. <P>COPYRIGHT: (C)2013,JPO&INPIT |
