Character recognition system which cuts between touching characters

• US 5,809,166 A
• Filed: 10/28/1997
• Issued: 09/15/1998
• Est. Priority Date: 11/09/1994
• Status: Expired due to Term

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Huang, Hung Khei, Yagasaki, Toshiaki
• Primary Examiner(s): PATEL, JAYANTI K
• Assistant Examiner(s): Bobys, Marc
• Application Number:
  US08/962,974
• Time in Patent Office: 322 Days
• Field of Search: 382/156, 382/171, 382/174, 382/177-179, 382/193, 382/196, 382/187
• US Class Current: 382/178
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/15 Cutting or merging image elements, e.g. region growing, watershed or clustering-based techniques

Abstract

An optical character recognition system cuts between touching characters. A first cut is made between touching characters based on white spaces, and the cut characters are subjected to character recognition processing. All characters not recognized are then cut again. A pair of adjacent vertical bars is detected in a vertical histogram of character image data of unrecognized characters, the vertical bars having a vertical component in the histogram that exceeds a predetermined vertical threshold. Horizontal crossings are detected in each of three discrete horizontal bands between the vertical bars. The vertical bars are classified according to the detected horizontal crossings, and, based on the classification, a decision is made whether or not to cut between the vertical bars, and where to cut between the vertical bars.

Claims

1. A method for cutting individual characters out of character image data which includes at least two touching characters, said method comprising the steps of:

detecting a first pair of vertical bars in the character image data;

checking for a presence of image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

classifying the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain image data;

measuring a distance between the first pair of vertical bars;

comparing the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified; and

determining whether and where to cut between the first pair of vertical bars based on at least one of the classification and the comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first pair of vertical bars, when it is determined to cut, the determining step determines to cut at one of only two locations between the first pair of vertical bars.

2. A method according to claim 1, further comprising the step of generating a vertical histogram of the character image data, and wherein the detecting step includes detecting the first pair of vertical bars in the vertical histogram.

3. A method according the claim 2, further comprising the step of storing a predetermined vertical threshold, and wherein the detecting step includes detecting a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

4. A method according to claim 1, wherein the checking step includes checking for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

5. A method according to claim 4, wherein the vertical line is located at a center point between the first pair of vertical bars.

6. A method according to claim 1, wherein the checking step includes checking for the presence of image data in each of three discrete horizontal bands.

7. A method according to claim 1, further comprising the step of selecting character image data for character cutting which extends between a predetermined character baseline and a predetermined character maximum height line.

8. A method according to claim 1, wherein each plural discrete horizontal band has a same height.

9. A method according to claim 1, wherein the classifying step includes classifying the first pair of vertical bars into one of four classes of vertical bars, and wherein the checking step includes checking for the presence of image data in each of three discrete horizontal bands.

10. A method according to claim 9, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

11. A method according to claim 10, further comprising the steps of:

cutting between the first pair of vertical bars when both a set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class; and

detecting a next vertical bar when the first pair of vertical bars is classified as third class, and when both the set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

12. A method according to claim 11, wherein the set of cutting conditions relate to whether the distance between the first pair of vertical bars is greater than the one of the plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

13. A method according to claim 1, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the center point and a second vertical bar in the first pair of vertical bars.

14. A method according to claim 1, further comprising the steps of:

storing the plural predetermined distances;

storing a plurality of preset characters; and

deciding whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the determining step includes determining whether and where to cut between the first pair of vertical bars based additionally on whether a portion of the first pair of vertical bars matches one of the preset characters.

15. A method according to claim 1, further comprising the steps of:

detecting a next vertical bar when it is determined not to cut between the first pair of vertical bars; and

repeating the method for a next pair of vertical bars.

16. An apparatus for cutting individual characters out of character image data which includes at least two touching characters, said apparatus comprising:

detecting means for detecting a first pair of vertical bars in the character image data;

checking means for checking for a presence of image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

classifying means for classifying the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain image data;

measuring means for measuring a distance between the first pair of vertical bars;

comparing means for comparing the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified; and

determining means for determining whether and where to cut between the first pair of vertical bars based on at least one of the classification and the comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first pair of vertical bars, when it is determined to cut, the determining means determines to cut at one of only two locations between the first pair of vertical bars.

17. An apparatus according to claim 16, further comprising generating means for generating a vertical histogram of the character image data, and wherein the detecting means detects the first pair of vertical bars in the vertical histogram.

18. An apparatus according the claim 17, further comprising storing means-for storing a predetermined vertical threshold, and wherein the detecting means detects a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

19. An apparatus according to claim 16, wherein the checking means checks for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

20. An apparatus according to claim 19, wherein the vertical line is located at a center point between the first pair of vertical bars.

21. An apparatus according to claim 16, wherein the checking means checks for a presence of image data in each of three discrete horizontal bands.

22. An apparatus according to claim 16, further comprising selecting means for selecting character image data for character cutting which extends between a predetermined character baseline and a predetermined character maximum height line.

23. An apparatus according to claim 16, wherein each plural discrete horizontal band has a same height.

24. An apparatus according to claim 16, wherein the classifying means classifies the first pair of vertical bars into one of four classes of vertical bars, and wherein the checking means checks for the presence of image data in each of three discrete horizontal bands.

25. An apparatus according to claim 24, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

26. An apparatus according to claim 25, further comprising:

cutting means for cutting between the first pair of vertical bars when both a set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class; and

next bar detecting means for detecting a next vertical bar when the first pair of vertical bars is classified as third class, and when both the set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

27. An apparatus according to claim 26, wherein the set of cutting conditions relate to whether the distance between the first pair of vertical bars is greater than the one of the plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

28. An apparatus according to claim 16, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the center point and a second vertical bar in the first pair of vertical bars.

29. An apparatus according to claim 16, further comprising:

distance storing means for storing the plural predetermined distances;

character storing means for storing a plurality of preset characters; and

character judging means for judging whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the determining means determines whether and where to cut between the first pair of vertical bars based additionally on the output of the character judging means.

30. An apparatus according to claim 16, further comprising:

next bar detecting means for detecting a next vertical bar when it is determined not to cut between the first pair of vertical bars.

31. An optical character recognition method for recognizing characters in input image data, comprising the steps of:

inputting image data including character image data which represents images of characters;

cutting character image data out of the image data;

performing optical character recognition processing on the cut character image data;

selecting unrecognized cut character image data;

detecting a first pair of vertical bars in the unrecognized cut character image data;

checking for a presence of character image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

classifying the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain character image data;

measuring a distance between the first pair of vertical bars;

comparing the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified;

determining whether and where to cut between the first pair of vertical bars in the unrecognized cut character image data based on at least one of the classification and the comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first pair of vertical bars, when it is determined to cut, the determining step determines to cut at one of only two locations between the first pair of vertical bars;

cutting between the first pair of vertical bars in the unrecognized character image data when it is determined to cut between the vertical bars;

performing optical character recognition processing on newly-cut character image data; and

outputting recognized characters.

32. A method according to claim 31, further comprising the step of generating a vertical histogram of the unrecognized cut character image data, and wherein the detecting step includes detecting the first pair of vertical bars in the vertical histogram.

33. A method according the claim 32, further comprising the step of storing a predetermined vertical threshold, and wherein the detecting step includes detecting a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

34. A method according to claim 31, wherein the checking step includes checking for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

35. A method according to claim 34, wherein the vertical line is located at a center point between the first pair of vertical bars.

36. A method according to claim 31, wherein the checking step includes checking for a presence of image data in each of three discrete horizontal bands.

37. A method according to claim 31, wherein the selecting step includes selecting unrecognized cut character image data for character cutting which extends between a predetermined character baseline and a predetermined character maximum height line.

38. A method according to claim 31, wherein each plural discrete horizontal band has a same height.

39. A method according to claim 31, wherein the classifying step includes classifying the first pair of vertical bars into one of four classes of vertical bars, and wherein the checking step includes checking for the presence of image data in each of three discrete horizontal bands.

40. A method according to claim 39, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

41. A method according to claim 40, further comprising the step of:

detecting a next vertical bar when the first pair of vertical bars is classified as third class, and when both a set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class, and wherein the step of cutting between the first pair of vertical bars includes cutting between the first pair of vertical bars when both the set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

42. A method according to claim 41, wherein the set of cutting conditions relate to whether the distance between the first pair of vertical bars is greater than the one of the plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

43. A method according to claim 31, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the center point and a second vertical bar in the first pair of vertical bars.

44. A method according to claim 31, further comprising the steps of:

storing the plural predetermined distances;

storing a plurality of preset characters; and

deciding whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the determining step includes determining whether and where to cut between the first pair of vertical bars based additionally on whether a portion of the first pair of vertical bars matches one of the preset characters.

45. A method according to claim 31, further comprising the steps of:

detecting a next vertical bar when it is determined not to cut between the first pair of vertical bars; and

repeating the method for a next pair of vertical bars.

46. An optical character recognition apparatus for recognizing characters in input image data, comprising:

inputting means for inputting image data including character image data which represents images of characters;

cutting means for cutting character image data in the image data;

optical character recognition processing means for performing optical character recognition processing on the cut character image data;

selecting means for selecting unrecognized cut character image data;

detecting means for detecting a first pair of vertical bars in the unrecognized cut character image data;

checking means for checking for a presence of character image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

classifying means for classifying the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain character image data;

measuring means for measuring a distance between the first pair of vertical bars;

comparing means for comparing the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified;

determining means for determining whether and where to cut between the first pair of vertical bars in the unrecognized cut character image data based on at least one of the classification and the-comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first pair of vertical bars, when it is determined to cut, the determining means determines to cut at one of only two locations between the first pair of vertical bars;

bar cutting means for cutting between the first pair of vertical bars in the unrecognized character image data when it is determined to cut between the first pair of vertical bars;

optical character recognition processing means for performing optical character recognition processing on newly-cut character image data; and

outputting means for outputting recognized characters.

47. An apparatus according to claim 46, further comprising generating means for generating a vertical histogram of the unrecognized cut character image data, and wherein the detecting means detects the first pair of vertical bars in the vertical histogram.

48. An apparatus according the claim 47, further comprising storing means for storing a predetermined vertical threshold, and wherein the detecting means detects a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

49. An apparatus according to claim 46, wherein the checking means checks for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

50. An apparatus according to claim 49, wherein the vertical line is located at a center point between the first pair of vertical bars.

51. An apparatus according to claim 46, wherein the checking means checks for the presence of image data in each of three discrete horizontal bands.

52. An apparatus according to claim 46, wherein the selecting means selects unrecognized cut character image data for character cutting which extends between a predetermined character baseline and a predetermined character maximum height line.

53. An apparatus according to claim 46, wherein each plural discrete horizontal band has a same height.

54. An apparatus according to claim 46, wherein the classifying means classifies the first pair of vertical bars into one of four classes of vertical bars, and wherein the checking means checks for the presence of image data in each of three discrete horizontal bands.

55. An apparatus according to claim 54, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

56. An apparatus according to claim 55, further comprising:

next bar detecting means for detecting a next vertical bar when the first pair of vertical bars is classified as third class, and when both a set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class, and wherein the bar cutting means cuts between the first pair of vertical bars when both the set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

57. An apparatus according to claim 56, wherein the set of cutting conditions relate to whether the distance between the first pair of vertical bars is greater than the one of the plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

58. An apparatus according to claim 46, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the center point and a second vertical bar in the first pair of vertical bars.

59. An apparatus according to claim 46, further comprising:

distance storing means for storing the plural predetermined distances;

character storing means for storing a plurality of preset characters; and

character judging means for judging whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the determining means determines whether and where to cut between the first pair of vertical bars based additionally on the output of the character judging means.

60. An apparatus according to claim 46, further comprising:

next bar detecting means for detecting a next vertical bar when it is determined not to cut between the first pair of vertical bars.

61. Computer-executable process steps stored in a computer-readable medium, the computer-executable process steps to cut individual characters out of character image data which includes at least two touching characters, the computer-executable process steps comprising:

code to detect a first pair of vertical bars in the character image data;

code to check for a presence of image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

code to classify the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain image data;

code to measure a distance between the first pair of vertical bars;

code to compare the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified; and

code to determine whether and where to cut between the first pair of vertical bars based on at least one of the classification and the comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first Pair of vertical bars, when it is determined to cut, the code to determine determines to cut at one of only two locations between the first pair of vertical bars.

62. Computer-executable process steps according to claim 61, further comprising code to generate a vertical histogram of the character image data, and wherein the code to detect detects the first pair of vertical bars in the vertical histogram.

63. Computer-executable process steps according the claim 62, further comprising code to store a predetermined vertical threshold, and wherein the code to detect detects a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

64. Computer-executable process steps according to claim 61, wherein the code to check checks for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

65. Computer-executable process steps according to claim 64, wherein the vertical line is located at a center point between the first pair of vertical bars.

66. Computer-executable process steps according to claim 61, wherein the code to check checks for the presence of image data in each of three discrete horizontal bands.

67. Computer-executable process steps according to claim 61, further comprising code to select character image data for character cutting which extends between a predetermined character baseline and a predetermined character maximum height line.

68. Computer-executable process steps according to claim 61, wherein each plural discrete horizontal band has a same height.

69. Computer-executable process steps according to claim 61, wherein the code to classify classifies the first pair of vertical bars into one of four classes of vertical bars, and wherein the code to check checks for the presence of image data in each of three discrete horizontal bands.

70. Computer-executable process steps according to claim 69, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

71. Computer-executable process steps according to claim 70, further comprising:

code to cut between the first pair of vertical bars when both a set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class; and

code to detect a next vertical bar when the first pair of vertical bars is classified as third class, and when both the set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

72. Computer-executable process steps according to claim 71, wherein the set of cutting conditions relates to whether the distance between the first pair of vertical bars is greater than one of plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

73. Computer-executable process steps according to claim 61, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the center point and a second vertical bar in the first pair of vertical bars.

74. Computer-executable process steps according to claim 61, further comprising:

code to store the plural predetermined distances;

code to store a plurality of preset characters; and

code to decide whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the code to determine determines whether and where to cut between the first pair of vertical bars based additionally on whether a portion of the first pair of vertical bars matches one of the preset characters.

75. Computer-executable process steps according to claim 61, further comprising:

code to detect a next vertical bar when it is determined not to cut between the first pair of vertical bars; and

code to re-execute the process steps for a next pair of vertical bars.

76. Computer-executable process steps stored in a computer-readable medium, the computer-executable process steps comprising an optical character recognition method to recognize characters in input image data, the computer-executable process steps comprising:

code to input image data including character image data which represents images of characters;

code to cut character image data out of the image data;

code to perform optical character recognition processing on the cut character image data;

code to select unrecognized cut character image data;

code to detect a first pair of vertical bars in the unrecognized cut character image data;

code to check for a presence of character image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

code to classify the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain character image data;

code to measure a distance between the first pair of vertical bars;

code to compare the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified;

code to determine whether and where to cut between the first pair of vertical bars in the unrecognized cut character image data based on at least one of the classification and the comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first pair of vertical bars, the code to determine determines to cut at one of only two locations between the first pair of vertical bars;

code to cut between the first pair of vertical bars in the unrecognized character image data when it is determined to cut between the vertical bars;

code to perform optical character recognition processing on newly-cut character image data; and

code to output recognized characters.

77. Computer-executable process steps according to claim 76, further comprising code to generate a vertical histogram of the unrecognized cut character image data, and wherein the code to detect detects the first pair of vertical bars in the vertical histogram.

78. Computer-executable process steps according the claim 77, further comprising code to store a predetermined vertical threshold, and wherein the code to detect detects a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

79. Computer-executable process steps according to claim 76, wherein the code to check checks for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

80. Computer-executable process steps according to claim 79, wherein the vertical line is located at a center point between the first pair of vertical bars.

81. Computer-executable process steps according to claim 76, wherein code to check checks for a presence of image data in each of three discrete horizontal bands.

82. Computer-executable process steps according to claim 76, wherein the code to select selects unrecognized cut character image data for character cutting which extends between a predetermined character baseline and a predetermined character maximum height line.

83. Computer-executable process steps according to claim 76, wherein each plural discrete horizontal band has a same height.

84. Computer-executable process steps according to claim 76, wherein the code to classify classifies the first pair of vertical bars into one of four classes of vertical bars, and wherein the code to check checks for the presence of image data in each of three discrete horizontal bands.

85. Computer-executable process steps according to claim 84, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

86. Computer-executable process steps according to claim 85, further comprising:

code to detect a next vertical bar when the first pair of vertical bars is classified as third class, and when both a set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class, and wherein the code to cut cuts between the first pair of vertical bars when both the set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

87. Computer-executable process steps according to claim 86, wherein the set of cutting conditions relates to whether the distance between the first pair of vertical bars is greater than the one of the plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

88. Computer-executable process steps according to claim 76, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the center point and a second vertical bar in the first pair of vertical bars.

89. Computer-executable process steps according to claim 76, further comprising:

code to store a plurality of predetermined distance conditions;

code to store a plurality of preset characters; and

code to decide whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the code to determine determines whether and where to cut between the first pair of vertical bars based additionally on whether a portion of the first pair of vertical bars matches one of the preset characters.

90. Computer-executable process steps according to claim 76, further comprising:

code to detect a next vertical bar when it is determined not to cut between the first pair of vertical bars; and

code to re-execute the process steps for a next pair of vertical bars.

91. A computer-readable memory medium which stores computer-executable process steps to cut individual characters out of character image data which includes at least two touching characters, the computer-executable process steps comprising:

a detecting step to detect a first pair of vertical bars in the character image data;

a checking step to check for a presence of image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

a classifying step to classify the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain image data;

a measuring step to measure a distance between the first pair of vertical bars;

a comparing step to compare the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified; and

a determining step to determine whether and where to cut between the first pair of vertical bars based on at least one of the classification and the comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first pair of vertical bars, when it is determined to cut, the determining step determines to cut at one of only two locations between the first pair of vertical bars.

92. A computer-readable memory medium according to claim 91, further comprising a generating step to generate a vertical histogram of the character image data, and wherein the detecting step detects the first pair of vertical bars in the vertical histogram.

93. A computer-readable memory medium according the claim 92, further comprising a storing step to store a predetermined vertical threshold, and wherein the detecting step detects a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

94. A computer-readable memory medium according to claim 91, wherein the checking step checks for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

95. A computer-readable memory medium according to claim 94, wherein the vertical line is located at a center point between the first pair of vertical bars.

96. A computer-readable memory medium according to claim 91, wherein the checking step checks for the presence of image data in each of three discrete horizontal bands.

97. A computer-readable memory medium according to claim 91, further comprising a selecting step to select character image data for character cutting which extends between a predetermined character baseline and a predetermined character maximum height line.

98. A computer-readable memory medium according to claim 91, wherein each plural discrete horizontal band has a same height.

99. A computer-readable memory medium according to claim 91, wherein the classifying step classifies the first pair of vertical bars into one of four classes of vertical bars, and wherein the checking step checks for the presence of image data in each of three discrete horizontal bands.

100. A computer-readable memory medium according to claim 99, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

101. A computer-readable memory medium according to claim 100, further comprising:

a cutting step to cut between the first pair of vertical bars when both a set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class; and

a detecting step to detect a next vertical bar when the first pair of vertical bars is classified as third class, and when both the set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

102. A computer-readable memory medium according to claim 101, wherein the set of cutting conditions relates to whether the distance between the first pair of vertical bars is greater than one of plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

103. A computer-readable memory medium according to claim 91, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the center point and a second vertical bar in the first pair of vertical bars.

104. A computer-readable memory medium according to claim 91, further comprising:

a storing step to store the plural predetermined distances;

a second storing step to store a plurality of preset characters; and

a deciding step to decide whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the determining step determines whether and where to cut between the first pair of vertical bars based additionally whether a portion of the first pair of vertical bars matches one of the preset characters.

105. A computer-readable memory medium according to claim 91, further comprising:

a detecting step to detect a next vertical bar when it is determined not to cut between the first pair of vertical bars; and

a re-executing step to re-execute the process steps for a next pair of vertical bars.

106. A computer-readable memory medium which stores computer-executable process steps to execute an optical character recognition method to recognize characters in input image data, the computer-executable process steps comprising:

an inputting step to input image data including character image data which represents images of characters;

a first cutting step to cut character image data out of the image data;

a first optical character recognition processing step to perform optical character recognition processing on the cut character image data;

a selecting step to select unrecognized cut character image data;

a detecting step to detect a first pair of vertical bars in the unrecognized cut character image data;

a checking step to check for a presence of character image data in each of plural discrete horizontal bands located between the first pair of vertical bars;

a classifying step to classify the first pair of vertical bars into one of plural classes based on which of the plural discrete horizontal bands contain character image data;

a measuring step to measure a distance between the first pair of vertical bars;

a comparing step to compare the distance between the first pair of vertical bars to one of plural predetermined distances, the one of the plural predetermined distances corresponding to the class into which the first pair of vertical bars has been classified;

a determining step to determine whether and where to cut between the first pair of vertical bars in the unrecognized cut character image data based on at least one of the classification and the comparison between the first pair of vertical bars and the one of the plural predetermined distances, wherein, for all classifications of the first pair of vertical bars, when it is determined to cut, the determining step determines to cut at one of only two locations between the first pair of vertical bars;

a second cutting step to cut between the first pair of vertical bars in the unrecognized character image data when it is determined to cut between the vertical bars;

a second optical character recognition processing step to perform optical character recognition processing on newly-cut character image data; and

an outputting step to output recognized characters.

107. A computer-readable memory medium according to claim 106, further comprising a generating step to generate a vertical histogram of the unrecognized cut character image data, and wherein the detecting step detects the first pair of vertical bars in the vertical histogram.

108. A computer-readable memory medium according the claim 107, further comprising a storing step to store a predetermined vertical threshold, and wherein the detecting step detects a first pair of vertical bars comprised of two vertical bars each having a vertical component that exceeds the predetermined vertical threshold.

109. A computer-readable memory medium according to claim 106, wherein the checking step checks for the presence of image data by checking for horizontal crossings on a vertical line located between the first pair of vertical bars and passing through each horizontal band.

110. A computer-readable memory medium according to claim 109, wherein the vertical line is located at a center point between the first pair of vertical bars.

111. A computer-readable memory medium according to claim 106, wherein the checking step checks for a presence of image data in each of three discrete horizontal bands.

112. A computer-readable memory medium according to claim 106, wherein the selecting step selects unrecognized cut character image data for character cutting which extends between a predetermined character base line and a predetermined character maximum height line.

113. A computer-readable memory medium according to claim 106, wherein each plural discrete horizontal band has a same height.

114. A computer-readable memory medium according to claim 106, wherein the classifying step classifies the first pair of vertical bars into one of four classes of vertical bars, and wherein the checking step checks for the presence of image data in each of three discrete horizontal bands.

115. A computer-readable memory medium according to claim 114, wherein the four classes of vertical bars include

(i) a first class of vertical bars in which (a) image data is present in only both a first horizontal band and a third horizontal band, (b) image data is present in only the first horizontal band, and (c) image data is absent from all three horizontal bands,

(ii) a second class of vertical bars in which (a) image data is present in only both a second horizontal band and the third horizontal band, and (b) image data is present in all three horizontal bands,

(iii) a third class of vertical bars in which (a) image data is present in only the second horizontal band, and (b) image data is present in only both the first horizontal band and the second horizontal band, and

(iv) a fourth class of vertical bars in which image data is present in only the third horizontal band.

116. A computer-readable memory medium according to claim 115, further comprising a detecting step to detect a next vertical bar when the first pair of vertical bars is classified as third class, and when both a set of cutting conditions is unsatisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class, and wherein the cutting step cuts between the first pair of vertical bars when both the set of cutting conditions is satisfied and the first pair of vertical bars is classified as one of first class, second class, and fourth class.

117. A computer-readable memory medium according to claim 116, wherein the set of cutting conditions relates to whether the distance between the first pair of vertical barks is greater than the one of the plural predetermined distances and whether the first pair of vertical bars represents a predetermined character.

118. A computer-readable memory medium according to claim 106, wherein the two points include a center point located an equal distance from each vertical bar in the first pair of vertical bars, and an offset point located between the-center point and a second vertical bar in the first pair of vertical bars.

119. A computer-readable memory medium according to claim 106, further comprising:

a storing step to store a plurality of predetermined distance conditions;

a second-storing step to store a plurality of preset characters; and

a deciding step to decide whether a portion of the first pair of vertical bars matches one of the preset characters;

wherein the determining step determines whether and where to cut between the first pair of vertical bars based additionally on whether a portion of the first pair of vertical bars matches one of the preset characters.

120. A computer-readable memory medium according to claim 106, further comprising:

a detecting step to detect a next vertical bar when it is determined not to cut between the first pair of vertical bars; and

a re-executing step to re-execute the process steps for a next pair of vertical bars.