Method of a system for analyzing characters

• US 4,811,412 A
• Filed: 01/11/1988
• Issued: 03/07/1989
• Est. Priority Date: 01/26/1987
• Status: Expired due to Term

• Current Assignee: Sharp Kabushiki Kaisha (Hon Hai Precision Industry Co., Ltd.)
• Original Assignee: Sharp Kabushiki Kaisha (Hon Hai Precision Industry Co., Ltd.)
• Inventors: Katsurada, Morihiro
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): NOT, DEFINED
• Application Number:
  US07/142,867
• Time in Patent Office: 421 Days
• Field of Search: 382/9, 382/47, 382/57
• US Class Current: 382/179
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/146 Aligning or centring of the image pick-up or image-field

Abstract

After an optical character reader scans a line and image data corresponding thereto are stored, vacant regions perpendicular to the direction of the line are identified. Thereafter, candidate character positions are identified firstly by starting from the beginning of the line and secondly by starting from the end thereof. If the results of the two analyses coincide, they are identified as character positions where neighboring characters are separated. If they do not, a pattern matching method is used with stored patterns at each candidate character position to determine whether it really represents a character position.

Claims

1. In a system for analyzing characters by extracting line image data from character image data in an image memory, storing said extracted line image data in a line image buffer memory, extracting individual characters in a line from said line image data, and identifying characters by matching characteristics from one of said extracted individual characters with preliminarily stored characteristics data, the improvement wherein said system further comprises:

candidate position detection means for analyzing said line image data stored in said line image buffer memory to determine whether character data are present or absent in a direction perpendicular to said line, thereby identifying starting and end positions of character data separated by vacant regions in said perpendicular direction and storing data indicative for said starting and end positions,

first character position judging means for specifying a starting position from the beginning of a line, identifying the farthest end position in said line from said specified starting position within a specified range, identifying the area between said specified starting position and said identified farthest end position as a candidate character position, and repeating this process to the end of said line,

second character position judging means for specifying an end position from the end of a line, identifying the farthest starting position in said line from said specified end position within a specified range, identifying the area between said identified farthest starting position and said specified end position as candidate character position, and repeating this process to the beginning of said line, and

character position determining means for identifying, if said candidate positions identified by said first and second character position judging means match, said candidate character positions identified by said first and second character position judging means as an individual character position and extracting characteristics at each of said candidate positions; and

identifying an individual character position by comparing said extracted characteristics with stored patterns in a dictionary memory if said candidate positions identified by said first and second character position judging means do not match.

2. A method of analyzing characters by extracting line image data from character image data in an image memory, storing said extracted line image data in a line image buffer means, extracting individual characters in a line from said line image data, and identifying characters by matching characteristics extracted from one of said extracted individual characters with preliminarily stored characteristics data, the improvement wherein said method further comprises the steps of:

analyzing said line image data stored in said line image buffer memory to determine whether character data are present or absent in a direction perpendicular to said line, thereby identifying starting and end positions of character data separated by vacant regions in said perpendicular direction and storing data indicative of said starting and end positions,

specifying a starting position from the beginning of a line, identifying the farthest end position in said line from said specified starting position within a specified range, identifying the area between said specified starting position and said identified farthest end position as one of first candidate character positions,

repeating the specifying and identifying steps described immediately above to the end of said line,

specifying an end position from the end of a line,

identifying the farthest starting position in said line from said specified end position within a specified range,

identifying the area between said identified farthest starting position and said specified end position as one of said candidate character positions,

repeating the specifying and identifying steps described immediately above to the beginning of said line, and

identifying, if said first and second candidate character positions match, said matching first and second candidate character positions as an individual character position and extracting characteristics at each of said first and second candidate character positions; and

identifying individual character positions by comparing said extracted characteristics with stored patterns in a dictionary memory if said first and second candidate character positions do not match.