Method and system of handwritten-character recognition

• US 6,289,124 B1
• Filed: 04/26/1999
• Issued: 09/11/2001
• Est. Priority Date: 04/27/1998
• Status: Expired due to Fees

• Current Assignee: Panasonic Healthcare Company Limited (PHC Holdings Corporation)
• Original Assignee: Sanyo Electric Company Limited (Panasonic Holdings Corporation)
• Inventors: Okamoto, Masayoshi
• Primary Examiner(s): Bella, Matthew C.
• Assistant Examiner(s): DESIRE, GREGORY M
• Application Number:
  US09/298,989
• Time in Patent Office: 869 Days
• Field of Search: 382/181-190, 382/192, 382/195, 382/197, 382/201, 382/202, 382/206, 382/209, 382/217-219, 382/224
• US Class Current: 382/187
• 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 30/1423 the instrument generating sequences of position coordinates corresponding to handwriting preprocessing or recognising digital ink G06V30/32
• 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

In accordance with the present character recognition method, an imaginary stroke is used to link from the ending point of each actual stroke of an input handwritten character to the starting point of the subsequent actual stroke thereof to form a single line. Then a feature level is detected for specifying the position of a turn of the single line and the direction and angle of the turn at the position. According to the detected position of the turn, the detected feature level is patterned on input mesh memories which are in turn compared with a previously formed dictionary mesh memory to calculate the resemblance of the input handwritten character to each handwritten character in a dictionary database. The handwritten character in the dictionary database that has the closest, calculated resemblance is recognized as the input handwritten character. According to the present method, an imaginary stroke added to an input handwritten character also allows correct recognition of a character with each stroke written cursively.

Claims

1. A method of recognizing an input handwritten character, comprising the steps of:

linking from an ending point of each actual stroke of said input handwritten character to a starting point of an actual stroke subsequent to said each actual stroke with an imaginary stroke to form a line;

detecting a feature level for specifying a turn of said line;

comparing said detected feature level with each feature level of a plurality of handwritten characters in a previously formed dictionary database to calculate a resemblance of said input handwritten character to each said handwritten character in said dictionary database; and

determining as said input handwritten character a handwritten character in said dictionary database having said calculated resemblance closest to said input handwritten character.

2. The method of claim 1, wherein said feature level includes a position of said turn.

3. The method of claim 2, wherein said feature level includes at least one of a direction of said turn and an angle of said turn.

4. The method of claim 3, wherein said step of detecting includes the step of identifying said direction of said turn depending on a plurality of predetermined reference directions.

5. The method of claim 4, wherein said step of identifying includes the step of assuming as said direction of said turn one of said plurality of reference directions closest to said direction of said turn.

6. The method of claim 4, wherein said step of identifying includes the step of assuming as said direction of said turn both two of said plurality of reference directions immediately adjacent to said direction of said turn.

7. The method of claim 6, wherein said step of identifying further includes the step of weighting said angle of said turn depending on a proximity of each of said both two reference directions to said direction of said turn.

8. The method of claim 2, further comprising the step of blurring said detected position of said turn after said step of detecting and before said step of comparing.

9. The method of claim 8, wherein said step of blurring includes the step of providing a highest score to the said detected position of said turn and a lower score to a position more distant from said detected position of said turn.

10. The method of claim 1, wherein each feature level of said handwritten characters in said dictionary database is calculated by averaging feature levels of a plurality of handwritten character samples by the number of said samples.

11. The method of claim 1, further comprising the step of using said detected feature level of said input handwritten character to update a feature level of said determined handwritten character in said dictionary database.

12. The method of claim 1, wherein said step of comparing compares said detected feature level with each feature level of said handwritten characters in said dictionary database without discriminating between said feature level for specifying said turn of said actual stroke and a feature level for specifying a turn between said actual stroke and said imaginary stroke.

13. The method of claim 1, wherein said step of comparing compares said detected feature level with each feature level of said handwritten characters in said dictionary database after weighting said feature level for specifying said turn of said actual stroke with a first coefficient and weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke with a second coefficient differing from said first coefficient.

14. The method of claim 13, wherein said second coefficient is smaller than said first coefficient.

15. The method of claim 1, wherein said step of comparing compares said detected feature level with each feature level of said handwritten characters in said dictionary database after weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke having a first length with a fast coefficient and weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke having a second length longer than said first length with a second coefficient smaller than said first coefficient.

16. The method of claim 1, further comprising the step of smoothing a portion linking said actual stroke and said imaginary stroke together.

17. The method of claim 16, wherein the step of smoothing repositions a starting point and/or an ending point of said actual stroke at a position closer to said actual stroke and an imaginary stroke linked to said actual stroke starting point or ending point.

18. The method of claim 17, wherein said step of smoothing includes the steps of:

selecting a point on said actual stroke detected earlier and/or later than said actual stroke starting point and/or ending point by a predetermined period of time to calculate a distance from said actual stroke starting point and/or ending point to said point selected on said actual stroke;

detecting a point on said imaginary stroke distant by said calculated distance from said imaginary stroke ending point and/or starting point linked to said actual stroke starting point and/or ending point; and

using said point selected on said actual stroke and said point detected on said imaginary stroke to determine said closer position.

19. The method of claim 17, said step of smoothing determines said closer position depending on a position of a point on said actual stroke distant from said actual stroke starting point and/or ending point by a predetermined distance and a point on said imaginary stroke distant by said predetermined distance from said imaginary stroke ending point and/or starting point linked to said actual stroke starting point or ending point.

20. The method of claim 16, further comprising the step of extending said actual stroke and adding an imaginary starting point and an imaginary ending point, wherein:

said step of linking links from said imaginary ending point of each actual stroke of said input handwritten character to said imaginary starting point of an actual stroke subsequent to said each actual stroke of said input handwritten character with imaginary stroke to form said line; and

said step of smoothing repositions said imaginary starting point and said imaginary ending point at a position closer to said actual stroke and an imaginary stroke linked to said imaginary starting point or imaginary ending point of said actual stroke.

21. A system for recognizing an input handwritten character, comprising:

a linking device constituted to link from an ending point of each actual stroke of said input handwritten character to a starting point of an actual stroke subsequent to said each actual stroke with an imaginary stroke to form a line;

a detecting device constituted to detect a feature level for specifying a turn of said line formed by said linking device;

a dictionary memory storing feature levels of a plurality of handwritten characters of a previously formed dictionary database;

a comparing device constituted to compare a feature level detected by said detecting device with each feature level of said plurality of handwritten characters of said dictionary database in said dictionary memory to calculate a resemblance of said input handwritten character to each said handwritten character in said dictionary database; and

a determining device constituted to determine as said input handwritten character a handwritten character in said dictionary database having the closest resemblance calculated by said comparing device.

22. The system of claim 21, wherein said feature level includes a position of said turn.

23. The system of claim 22, wherein said feature level includes at least one of a direction of said turn and an angle of said turn.

24. The system of claim 23, further comprising a temporary memory including a plurality of spatial, input-feature storage regions associated with to a plurality of predetermined reference directions, for storing a feature level in directional change for specifying said angle of said turn, wherein:

said dictionary memory includes a plurality of spatial, dictionary database feature storage regions associated with said plurality of reference directions, for storing a feature level in directional change for specifying said angle of said turn of said handwritten character in said dictionary database;

said detecting device stores said feature level in directional change into said temporary memory at an address corresponding to said position of said turn detected; and

said comparing device compares each of said spatial, input-feature storage regions in said temporary memory with an associated one of said spatial, dictionary database feature storage regions in said dictionary memory.

25. The system of claim 24, wherein said detecting device assumes as said direction of said turn one of said plurality of reference directions closest to said direction of said turn.

26. The system of claim 24, wherein said detecting device assumes as said direction of said turn both two of said plurality of reference directions immediately adjacent to said direction of said turn.

27. The system of claim 26, wherein said detecting device uses a proximity of each of said both two reference directions to said direction of said turn to weight said feature level in directional change and said detecting device stores the weighted said feature level in directional change into spatial, input-feature storage regions respectively associated with said both two reference directions.

28. The system of claim 22, further comprising a blurring device constituted to blur said position of said turn detected by said detecting device, before providing said position of said turn to said comparing device.

29. The system of claim 28, wherein said blurring device provides a highest score to said position of said turn detected by said detecting device and a lower score to a position more distant from said position of said turn detected by said detecting device.

30. The system of claim 21, wherein each feature level of said handwritten characters of said dictionary database in said dictionary memory is calculated by averaging feature levels of a plurality of handwritten character samples by the number of said samples.

31. The system of claim 21, further comprising an updating device constituted to update said feature level of said handwritten character in said dictionary database determined by said determining device, using said feature level of said input handwritten character detected by said detecting device.

32. The system of claim 31, wherein said comparing device compares said detected feature level with each feature level of said handwritten characters in said dictionary database without discriminating between said feature level for specifying said turn of said actual stroke and a feature level for specifying a turn between said actual stroke and said imaginary stroke.

33. The system of claim 31, wherein said comparing device compares said detected feature level with each feature level of said handwritten characters in said dictionary database after weighting said feature level for specifying said turn of said actual stroke with a first coefficient and weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke with a second coefficient differing from said first coefficient.

34. The system of claim 33, wherein said second coefficient is smaller than said first coefficient.

35. The system of claim 31, wherein said comparing device compares said detected feature level with each feature level of said handwritten characters in said dictionary database after weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke having a first length with a first coefficient and weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke having a second length longer than said first length with a second coefficient smaller than said first coefficient.

36. The system of claim 21, further comprising a smoothing device constituted to smooth a portion linking said actual stroke and said imaginary stroke together.

37. The system of claim 36, wherein said smoothing device repositions a starting point and/or an ending point of said actual stroke at a position closer to said actual stroke and an imaginary stroke linked to said actual stroke starting point or ending point.

38. The system of claim 37, wherein said smoothing device includes:

means for selecting a point on said actual stroke detected earlier and/or later than said actual stroke starting point and/or ending point by a predetermined period of time to calculate a distance from said actual stroke starting point and/or ending point to said point selected on said actual stroke;

means for detecting a point on said imaginary stroke distant by said calculated distance from said imaginary stroke ending point and/or starting point linked to said actual stroke starting point and/or ending point; and

mean for using said point selected on said actual stroke and said point detected on said imaginary stroke to determine said closer position.

39. The system of claim 37, said smoothing device determines said closer position depending on a position of a point on said actual stroke distant from said actual stroke starting point and/or ending point by a predetermined distance and a point on said imaginary stroke distant by said predetermined distance from said imaginary stroke ending point and/or starting point linked to said actual stroke starting point or ending point.

40. The system of claim 36, further comprising an extending device constituted to extend said actual stroke and add an imaginary starting point and an imaginary ending point, wherein:

said linking device links from said imaginary ending point of each actual stroke of said input handwritten character to said imaginary starting point of an actual stroke subsequent to said each actual stroke of said input handwritten character with an imaginary stroke to form said single line; and

said smoothing device repositions said imaginary starting point and said imaginary ending point at a position closer to said actual stroke and an imaginary stroke linked to said imaginary starting point or imaginary ending point of said actual stroke.

41. A medium storing a computer-readable program for recognizing an input handwritten character, said program allowing a computer to perform the steps of:

linking from an ending point of each actual stroke of said input handwritten character to a starting point of an actual stroke subsequent to said each actual stroke with an imaginary stroke to form a line;

detecting a feature level for specifying a turn of said line;

comparing said detected feature level with each feature level of a plurality of handwritten characters in a previously formed dictionary database to calculate a resemblance of said input handwritten character to each said handwritten character in said dictionary database; and

determining as said input handwritten character a handwritten character in said dictionary database having said calculated resemblance closest to said input handwritten character.

42. The medium of claim 41, wherein said feature level includes a position of said turn.

43. The medium of claim 42, wherein said feature level includes at least one of a direction of said turn and an angle of said turn.

44. The medium of claim 43, wherein said step of detecting includes the step of identifying said direction of said turn depending on a plurality of predetermined reference directions.

45. The medium of claim 44, wherein said step of identifying includes the step of assuming as said direction of said turn one of said plurality of reference directions closest to said direction of said turn.

46. The medium of claim 44, wherein said step of identifying includes the step of assuming as said direction of said turn both two of said plurality of reference directions immediately adjacent to said direction of said turn.

47. The medium of claim 46, wherein said step of identifying further includes the step of weighting said angle of said turn depending on a proximity of each of said both two reference directions to said direction of said turn.

48. The medium of claim 42, further comprising the step of blurring said detected position of said turn after said step of detecting and before said step of comparing.

49. The medium of claim 48, wherein said step of blurring includes the step of providing a highest score to the said detected position of said turn and a lower score to a position more distant from said detected position of said turn.

50. The medium of claim 41, wherein each feature level of said handwritten characters in said dictionary database is calculated by averaging feature levels of a plurality of handwritten character samples by the number of said samples.

51. The medium of claim 41, further comprising the step of using said detected feature level of said input handwritten character to update a feature level of said determined handwritten character in said dictionary database.

52. The medium of claim 51, wherein said step of comparing compares said detected feature level with each feature level of said handwritten characters in said dictionary database without discriminating between said feature level for specifying said turn of said actual stroke and a feature level for specifying a turn between said actual stroke and said imaginary stroke.

53. The medium of claim 51, wherein said step of comparing compares said detected feature level with each feature level of said handwritten characters in said dictionary database after weighting said feature level for specifying said turn of said actual stroke with a first coefficient and weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke with a second coefficient differing from said first coefficient.

54. The medium of claim 53, wherein said second coefficient is smaller than said first coefficient.

55. The medium of claim 51, wherein said step of comparing compares said detected feature level with each feature level of said handwritten characters in said dictionary database after weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke having a first length with a first coefficient and weighting a feature level for specifying a turn between said actual stroke and said imaginary stroke having a second length longer than said first length with a second coefficient smaller than said first coefficient.

56. The medium of claim 41, further comprising the step of smoothing a portion linking said actual stroke and said imaginary stroke together.

57. The medium of claim 56, wherein the step of smoothing repositions a starting point and/or an ending point of said actual stroke at a position closer to said actual stroke and an imaginary stroke linked to said actual stroke starting point or ending point.

58. The medium of claim 57, wherein said step of smoothing includes the steps of:

selecting a point on said actual stroke detected earlier and/or later than said actual stroke starting point and/or ending point by a predetermined period of time to calculate a distance from said actual stroke starting point and/or ending point to said point selected on said actual stroke;

detecting a point on said imaginary stroke distant by said calculated distance from said imaginary stroke ending point and/or starting point linked to said actual stroke starting point and/or ending point; and

using said point selected on said actual stroke and said point detected on said imaginary stroke to determine said closer position.

59. The medium of claim 57, said step of smoothing determines said closer position depending on a position of a point on said actual stroke distant from said actual stroke starting point and/or ending point by a predetermined distance and a point on said imaginary stroke distant by said predetermined distance from said imaginary stroke ending point and/or starting point linked to said actual stroke starting point or ending point.

60. The medium of claim 56, further comprising the step of extending said actual stroke and adding an imaginary starting point and an imaginary ending point, wherein:

said step of linking links from said imaginary ending point of each actual stroke of said input handwritten character to said imaginary starting point of an actual stroke subsequent to said each actual stroke of said input handwritten character with imaginary stroke to form said line; and

said step of smoothing repositions said imaginary starting point and said imaginary ending point at a position closer to said actual stroke and an imaginary stroke linked to said imaginary starting point or imaginary ending point of said actual stroke.