Title:
ENGLISH CHARACTER RECOGNITION METHOD
Document Type and Number:
Japanese Patent JP3151866
Kind Code:
B2
Abstract:
PURPOSE: To provide a Latin character recognition method in which working efficiency is remarkably improved by increasing the Latin character recognition rate in general documents and enabling the shortening of a processing time.
CONSTITUTION: After a character area is extracted from an input picture, a segment is performed from the extracted character area at a word unit and character candidate graphic within the word is further extracted, the segmented character candidate graphic is classified into some character pattern groups by a superposing method and a similarity degree method, etc., the recognition of the representative pattern of the classified character pattern groups is performed, an English word dictionary and the patterns are compared by defining a high recognition accuracy as a key. As for the character pattern group with low recognition accuracy, this method has a configuration that the recognition character is determined.
Inventors:
Noboru Nakamura
Application Number:
JP19311891A
Publication Date:
April 03, 2001
Filing Date:
August 01, 1991
Export Citation:
Assignee:
Matsushita Electric Industrial Co., Ltd
International Classes:
G06K9/72; G06K9/62; (IPC1-7): G06K9/72
Domestic Patent References:
| JP59149569A | ||||
| JP3186985A |
Other References:
糸乗，尾崎“類似文字による日本語単語抽出”信学技報（ＰＲＭＵ９８−８７），電子情報通信学会，ｐ２５−ｐ３２．１９９８年９月
Attorney, Agent or Firm:
Fumio Iwahashi (2 others)