Apparatus and method for recognizing character

• US 6,643,401 B1
• Filed: 06/24/1999
• Issued: 11/04/2003
• Est. Priority Date: 12/16/1998
• Status: Expired due to Fees

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Kashioka, Junji, Naoi, Satoshi
• Primary Examiner(s): Mariam, Daniel G.
• Application Number:
  US09/339,200
• Time in Patent Office: 1,594 Days
• Field of Search: 382/100, 382/101, 382/102, 382/159, 382/170, 382/176, 382/177, 382/178, 382/179, 382/181, 382/182, 382/185, 382/186, 382/187, 382/190-193, 382/195, 382/197, 382/198, 382/202, 382/209, 382/218, 382/224, 382/229, 382/294, 382/305, 209/584, 209/900, 345/191, 707/6, 707/7, 715/541
• US Class Current: 382/197
• CPC Class Codes: G06V 30/2445 Alphabet recognition, e.g. Latin, Kanji or Katakana
• CPC Class Codes: G06V 30/2455 Discrimination between machine-print, hand-print and cursive writing

Abstract

A character pattern is extracted from image data read from a document, listing, etc., and discriminated between a hand-written character and a typed character by a hand-written/typed character discrimination unit. The hand-written/typed character discrimination unit obtains, from the character pattern, N feature vectors containing a feature indicating at least the complexity and the linearity of the character pattern; and discriminating the character pattern between a hand-written character and a typed character using the feature vectors. A character recognition unit performs a character recognizing process based on the result of discriminating whether the character data is a hand-written character or a typed character. As a feature of the above described character pattern, the variance of line widths, the variance of character positions, etc. can also be used.

Claims

1. A method of recognizing a character, comprising:

extracting a character string pattern from an image in a document or a listing;

extracting a feature indicating complexity of the character string pattern;

extracting a feature indicating linearity of the character string pattern;

extracting a feature indicating a variance of character positions of the character string pattern;

obtaining N feature vectors containing a feature indicating at least complexity, linearity and a variance of character positions of the character string pattern; and

discriminating the character string pattern between a hand-written character and a typed character using the feature vectors, and performing a character recognizing process based on a discrimination result, wherein line density is used as a feature indicating the complexity.

2. A method of recognizing a character, comprising:

extracting a character string pattern from an image in a document or a listing;

extracting a feature amount indicating a variance of line widths obtained by removing from the character string pattern a pattern within a predetermined distance from an end point, an intersecting point, and a corner point of a character stroke contained in the character string pattern, and a feature amount indicating a variance of character positions of the character string pattern;

discriminating a hand-written character from a typed character based on at least the feature amount; and

recognizing a character based on a discrimination result, wherein the variance of line widths is obtained for each stroke or a stroke direction.

3. A character recognition apparatus comprising:

a character extraction unit extracting a character string pattern from an image in a document or a listing;

a hand-written/typed character discrimination unit discriminating the character string pattern between a hand-written character and a typed character; and

a character recognition unit recognizing a character based on a discrimination result, wherein

said hand-written/typed character discrimination unit comprises;

a complexity extraction unit extracting a feature indicating complexity of the character string pattern from the character string pattern;

a linearity extraction unit extracting a feature indicating linearity of the character string pattern;

a character position variance unit extracting a feature indicating a variance of character positions of the character string pattern;

a discrimination unit discriminating the character string pattern between a hand-written character and a typed character based on N feature vectors containing at least a feature indicating the complexity, a feature indicating the linearity and a variance of character positions of the character string pattern, wherein said complexity extraction unit extracts a feature indicating the complexity by obtaining line density from the character string pattern.

4. A character recognition apparatus comprising:

a character extraction unit extracting a character pattern from an image in a document or a listing;

a hand-written/typed character discrimination unit discriminating the character string pattern between a hand-written character and a typed character; and

a character recognition unit recognizing a character based on a discrimination result, wherein

said hand-written/typed character discrimination unit comprises;

a line width variance extraction unit extracting a feature amount indicating a variance of line widths obtained by removing from the character string pattern a pattern within a predetermined distance from an end point, an intersecting point, and a corner point of a character stroke contained in the character string pattern;

a character position variance unit extracting a feature indicating a variance of character positions of the character string pattern;

a discrimination unit discriminating the character string pattern between a hand-written character and a typed character based on at least a feature amount indicating the variance of line widths and a variance of character positions of the character string pattern, wherein said line width variance extraction unit obtains the variance of line widths for each stroke or stroke direction.

5. A character recognition apparatus, comprising:

a feature identification unit identifying a feature of a characterizing pattern extracted from an input image;

a hand-written character recognition unit character-recognizing a hand-written character;

a typed character recognition unit character-recognizing a typed character; and

a character recognition selection unit selecting one of said hand-written character recognition unit and said typed character recognition unit based on an identification result from said feature identification unit, wherein;

said feature identification unit discriminates the pattern between a hand-written character and a typed character based on regularity of a pattern;

a feature of the pattern contains at least one of complexity, linearity, variance of line widths, variance of arrangement positions, variance of tilts, variance of sizes, variance of character positions of the pattern;

said complexity contains at least one of line density, density of black picture elements, an Euler number, and an outline length; and

said linearity is computed based on a number of lines longer than a predetermined length in the pattern.

6. A computer-readable storage medium storing a character recognizing program for recognizing a character string pattern extracted from an image in a document or a listing, wherein

said character recognizing program directs a computer to perform operations of:

extracting a feature indicating complexity of the character string pattern;

extracting a feature indicating linearity of the character string pattern;

extracting a feature indicating a variance of character positions of the character string pattern;

obtaining, from the character string pattern, N feature vectors containing a feature indicating at least complexity, linearity and a variance of character positions of the character string pattern; and

discriminating the character string pattern between a hand-written character and a typed character using the feature vectors, wherein line density is used as a feature indicating the complexity.

7. A computer-readable storage medium storing a character recognizing program for recognizing a character string pattern extracted from an image in a document or a listing, wherein

said character recognizing program directs a computer to perform operations of:

extracting a feature amount indicating a variance of line widths obtained by removing from the character string pattern a pattern within a predetermined distance from an end point, an intersecting point, and a corner point of a character stroke contained in the character string pattern, and a feature amount indicating a variance of character positions of the character string pattern; and

discriminating a hand-written character from a typed character based on at least the feature amount, wherein said variance of line widths is obtained for each stroke or a stroke direction.

8. A character recognition method, comprising:

extracting a character string pattern from an image;

identifying a feature of the character string pattern extracted from the image; and

determining whether a character is a handwritten character or a typed character based on at least the identified feature, wherein said identifying comprises;

extracting a feature amount indicating a variance of line widths obtained by removing from the character string pattern a pattern within a predetermined distance from an end point, an intersecting point and a corner point of a character stroke contained in the character string pattern;

extracting a feature indicating complexity of the character string pattern;

extracting a feature indicating linearity of the character string pattern;

extracting a feature indicating a variance of character positions of the character string pattern;

obtaining feature vectors containing a feature indicating at least complexity, linearity and a variance of character positions of the character string pattern, and wherein said determining is performed by using the feature vectors.