| US 7,508,986 B2 | ||
| Document recognition device, document recognition method and program, and storage medium | ||
| Hiroaki Ikeda, Tokyo (Japan); Keiko Nakanishi, Tokyo (Japan); and Taeko Gotoh, Tokyo (Japan) | ||
| Assigned to Canon Kabushiki Kaisha, Tokyo (Japan) | ||
| Filed on Nov. 23, 2004, as Appl. No. 10/997,537. | ||
| Claims priority of application No. 2003-398425 (JP), filed on Nov. 28, 2003; application No. 2003-398426 (JP), filed on Nov. 28, 2003; and application No. 2004-080984 (JP), filed on Mar. 19, 2004. | ||
| Prior Publication US 2005/0117803 A1, Jun. 02, 2005 | ||
| Int. Cl. G06K 9/00 (2006.01); G06K 9/34 (2006.01); G06K 9/62 (2006.01); G06K 9/68 (2006.01) | ||
| U.S. Cl. 382—203 [382/162; 382/175; 382/209; 382/218; 382/219] | 10 Claims |

| 1. A document recognition device for recognizing a registered document, which is similar to an input document image, from
a plurality of registered documents, the document recognition device comprising:
shape-feature extracting means which extracts the shape feature from a component included in the document image;
color-information extracting means which extracts color information from the component included in the document image;
similarity-degree calculating means which calculates the degree of similarity of shape between the document image and the
registered document based on the shape feature extracted by the shape-feature extracting means;
similarity-degree correcting means which calculates the degree of similarity of color information extracted by the color-information
extracting means and corrects the degree of similarity calculated by the similarity-degree calculating means based on the
degree of similarity of color information;
determining means which determines whether the registered document is similar to the document image based on the degree of
similarity obtained by the similarity-degree correcting means; and
merging means which merges adjacent components among the components of the document image when the adjacent components have
the same color information,
wherein the similarity-degree calculating means calculates the degree of similarity of shape based on the components merged
by the merging means and the component which are not merged, and the similarity-degree correcting means calculates the degree
of similarity of color information and corrects the degree of similarity of shape.
|