Character recognition method and apparatus

• US 5,734,750 A
• Filed: 02/15/1995
• Issued: 03/31/1998
• Est. Priority Date: 10/16/1991
• Status: Expired due to Term

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Matsubayashi, Kazuhiro, Arai, Tsunekazu, Mori, Shigeki, Sakaguchi, Katsuhiko
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): MEHTA, BHAVESH M
• Application Number:
  US08/390,275
• Time in Patent Office: 1,140 Days
• Field of Search: 382/185, 382/187, 382/202, 382/119, 382/224, 382/209, 382/197, 382/186, 382/198, 382/203
• US Class Current: 382/202
• CPC Class Codes: G06V 10/20 Image preprocessing
• CPC Class Codes: G06V 10/44 Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/347 Sampling; Contour coding; Stroke extraction
• CPC Class Codes: G06V 30/373 using a special pattern or subpattern alphabet

Abstract

A character recognition apparatus and method for recognizing an unknown character by comparing strokes of the input character with stroke information stored in a dictionary. Each stroke of the input character is classified into one of plural categories according to stroke direction. The classified input strokes are rearranged into groups of strokes having the same classification. The degree of analogy between rearranged input strokes and stroke information stored in the dictionary for each group of similarly classified stroke group is calculated based on stroke length, and a maximum analogous element for each input stroke according to the calculated degree of analogy is detected. The detected maximum analogous elements are summed together, and the character whose sum is maximum is output as the identity of the unknown input character.

Claims

1. A character recognition apparatus comprising:

input means for inputting a plurality of strokes which collectively form a character;

input-stroke classification means for classifying each stroke of the input strokes, which are constituent strokes forming the input character, into one of a plurality of different types;

extraction means for extracting input strokes which are classified into one type;

calculating means for calculating a value of analogy between input strokes which are classified as a particular type and stroke information of the particular type for a character stored in a dictionary, for forming an analogy matrix comprising calculated values of analogy as elements thereof, for detecting a maximum matrix element in each of a plurality of columns or in each of a plurality of rows of the analogy matrix, and for designating the detected maximum matrix elements as maximum analogous elements;

determining means for determining whether or not there is one maximum matrix element in each row or each column of the analogy matrix, and in a case that there is more than one maximum matrix element in a row or a column, for selecting a maximum matrix element so that only one maximum matrix element is present in each row or each column;

detection means for detecting a maximum analogous character from a plurality of characters stored in the dictionary according to calculated values of analogy between input strokes classified as a particular type and stroke information of the particular type for each of the plurality of characters stored in the dictionary; and

output means for outputting said maximum analogous character.

2. A character recognition apparatus according to claim 1, wherein said input-stroke classification means classifies each stroke based on stroke direction.

3. A character recognition apparatus according to claim 1, further comprising stroke length determining means for determining stroke length of each stroke of the input character, and wherein said detection means calculates degree of analogy based on stroke length.

4. A character recognition apparatus according to claim 1, wherein the dictionary stroke information comprises strokes of a character pattern stored in the dictionary whose number of strokes coincides with the number of strokes of a pattern of the input character for each similarly classified stroke group.

5. A character recognition apparatus according to claim 4, wherein the dictionary has a data structure in which dictionary-stroke information of respective dictionary character patterns are arranged based on said input-stroke classification means.

6. A character recognition apparatus according to claim 1, wherein the dictionary has a data structure in which dictionary-stroke information of respective dictionary character patterns are arranged based on said input-stroke classification means.

7. A character recognition method comprising the steps of:

inputting a plurality of strokes which collectively form a character;

classifying each stroke of the input strokes into one of a plurality of different types;

extracting input strokes which are classified into one type;

calculating a value of analogy between the strokes which are classified as said type and stroke information of said type for a character stored in a dictionary;

forming an analogy matrix comprising calculated values of analogy as elements thereof;

detecting a maximum matrix element in each of a plurality of columns or in each of a plurality of rows of the analogy matrix;

designating the detected maximum matrix elements as maximum analogous elements;

determining whether or not there is one maximum matrix element in each row or each column of the analogy matrix;

selecting a maximum matrix element so that only one maximum matrix element is present in each row or each column, in a case that there is more than one maximum matrix element in a row or column;

detecting a maximum analogous character from a plurality of characters stored in the dictionary according to values of analogy between input strokes classified as a particular type and stroke information of the particular type for each of the plurality of characters stored in the dictionary; and

outputting said maximum analogous character.

8. A character recognition method according to claim 7, wherein said input-stroke classification step classifies each stroke based on stroke direction.

9. A character recognition method according to claim 7, further comprising the step of determining stroke length of each stroke of the input character, and wherein said calculating step calculates a value of analogy based on stroke length.

10. A character recognition method according to claim 7, wherein the dictionary stroke information comprises strokes of a character pattern stored in the dictionary whose number of strokes coincides with the number of strokes of a pattern of the input character for each similarly classified stroke group.

11. A character recognition apparatus according to claim 10, wherein the dictionary has a data structure in which dictionary-stroke information of respective dictionary character patterns are arranged based on said classifying step.

12. A character recognition method according to claim 7, wherein the dictionary has a data structure in which dictionary-stroke information of respective dictionary character patterns are arranged based on said classifying step.

13. A character recognition apparatus comprising:

input means for inputting a plurality of strokes constituting a character, the strokes being represented by a direction component and a coordinate position;

rearrangement means for rearranging the input strokes according to a sequence and a position of the input strokes;

calculating means for calculating a one-to-one correspondence value of analogy between input strokes rearranged by said rearrangement means and stroke information stored in the dictionary;

detection means for detecting a maximum analogous character according to the calculated values of analogy; and

output means for outputting said maximum analogous character.

14. A character recognition apparatus according to claim 13, further comprising stroke length determining means for determining stroke length of each stroke of the input character, and wherein said calculating means calculates a value of analogy based on stroke length.

15. A character recognition apparatus according to claim 13, further comprising means for determining whether or not there is one maximum analogous element corresponding to each input stroke, and, in a case that there is more than one maximum analogous element corresponding to an input stroke, for selecting an element so that only one maximum analogous element corresponds to each input stroke.

16. A character recognition apparatus according to claim 13, wherein the dictionary has a data structure in which dictionary-stroke information of respective dictionary character patterns are arranged based on said input-stroke classification means and said rearrangement means.

17. A character recognition method comprising the steps of:

inputting a plurality of strokes represented by a direction component and a coordinate position;

rearranging the input strokes according to a sequence and a position of the input strokes;

calculating a one-to-one correspondence value of analogy between input strokes rearranged in said rearrangement step and stroke information stored in the dictionary;

detecting a maximum analogous character according to the calculated values of analogy; and

outputting said maximum analogous character.

18. A character recognition method according to claim 17, further comprising the step of determining stroke length of each stroke of the input character, and calculating degree of analogy based on stroke length.

19. A character recognition method according to claim 17, further comprising the step of determining whether or not there is one maximum analogous element corresponding to each input stroke, and, in a case that there is more than one maximum analogous element corresponding to an input stroke, selecting an element so that only one maximum analogous element corresponds to each input stroke.

20. A character recognition method according to claim 17, wherein the dictionary has a data structure in which dictionary-stroke information of respective dictionary character patterns are arranged based on said input-stroke classification means and said rearrangement means.

21. A computer-readable medium having stored thereon computer-executable process steps, the computer-executable process steps comprising:

an inputting step to input a plurality of strokes which collectively form a character;

a classifying step to classify each stroke of the input strokes into one of a plurality of different types;

an extracting step to extract input strokes which are classified into one type;

a calculating step to calculate a value of analogy between the strokes which are classified as said type and stroke information of said type for a character stored in a dictionary;

a forming step to form an analogy matrix comprising calculated values of analogy as elements thereof;

a detecting step to detect a maximum matrix element in each of a plurality of columns or in each of a plurality of rows of the analogy matrix;

a designating step to designate the detected maximum matrix elements as maximum analogous elements;

a determining step to determine whether or not there is one maximum matrix element in each row or each column of the analogy matrix;

a selecting step to select a maximum matrix element so that only one maximum matrix element is present in each row or each column, in a case that there is more than one maximum matrix element in a row or column;

a detecting step to detect a maximum analogous character from a plurality of characters stored in the dictionary according to values of analogy between input strokes classified as a particular type and stroke information of the particular type for each of the plurality of characters stored in the dictionary; and

an outputting step to output said maximum analogous character.

22. A computer-readable medium having stored thereon computer-executable process steps, the computer-executable process steps comprising:

an inputting step to input a plurality of strokes represented by a direction component and a coordinate position;

a rearranging step to rearrange the input strokes according to a sequence and a position of the input strokes;

a calculating step to calculate a one-to-one correspondence value of analogy between rearranged input strokes and stroke information stored in the dictionary;

a detecting step to detect a maximum analogous character according to the calculated values of analogy; and

an outputting step to output said maximum analogous character.