Title:
【発明の名称】文字認識装置
Document Type and Number:
Japanese Patent JP2980636
Kind Code:
B2
Abstract:
PURPOSE:To absorb the variation of feature due to difference in the form of characters and to improve processing speed and recognition accuracy by respectively extracting the number of branches for respective picture elements from plural detecting branches and detecting points corresponding to the respective picture elements and limiting a character pattern. CONSTITUTION:The device is provided with a first branch extracting part 16 for extracting the first number of branches for respective picture elements from plural detecting branches corresponding to the respective picture elements existing in a character line area within a character pattern, and a second branch extracting part 17 for respectively extracting the second number of of branches for the respective picture elements from plural detecting points corresponding to the respective picture elements in the character line area. Further, a picture element processing part 19 for removing the picture element for which the number of these branches satisfies a prescribed condition from the character pattern is provided. Thus, the picture elements for which the first and the second number of branches satisfy the prescribed condition are removed from the character pattern. A portion more redundant than the standard character form of a feature dictionary is deleted. The variation of feature due to the difference of the character form can be absorbed by a simple method. Accordingly, processing speed and recognition accuracy are improved.
Inventors:
NAKAMURA TETSUO
HIGUCHI KOICHI
YAMASHITA YOSHUKI
HIGUCHI KOICHI
YAMASHITA YOSHUKI
Application Number:
JP8316290A
Publication Date:
November 22, 1999
Filing Date:
March 30, 1990
Export Citation:
Assignee:
OKI DENKI KOGYO KK
International Classes:
G06K9/46; G06K9/62; (IPC1-7): G06K9/46
Attorney, Agent or Firm:
Kakimoto Kyosei