System, process and software arrangement for recognizing handwritten characters

• US 7,336,827 B2
• Filed: 11/08/2001
• Issued: 02/26/2008
• Est. Priority Date: 11/08/2000
• Status: Expired due to Term

• Current Assignee: New York University
• Original Assignee: New York University
• Inventors: Pipes, Mark, Oh, Jong, Geiger, Davi
• Primary Examiner(s): AZARIAN, SEYED H
• Application Number:
  US10/416,319
• Publication Number:
  US 20060050962A1
• Time in Patent Office: 2,301 Days
• Field of Search: 382/106, 382112-114, 382/116, 382/119, 382/123, 382/137, 382/140, 382/168, 382/181, 382/202, 382184-189, 382/203, 382/209, 382/215, 382/231, 382/232, 382/254, 382/287, 382/290, 382/294, 382/301, 382/305, 382/314, 382173-179, 706/12, 706/6, 715/541
• US Class Current: 382/186
• 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 30/2272 with lexical matching

Abstract

A process and system are provided for determining a most likely combination of characters. In particular, character data which includes information indicative of at least one handwritten character is obtained. The character data includes at least one set of segmentation points for the handwritten character. Then, a score can be provided for each particular character of a set of previously stored characters based on a comparison between the character data and the previously stored particular character. In addition, it is possible to compare visual aspects of the handwritten character to visual aspects of each of the previously stored characters for determining likely characters. Also, a Fisher Matching procedure can be used on the character data to ascertain the likely characters. A plurality of handwritten characters can include a first character that is connected to a second character of the handwritten characters via a ligature. A further score of the ligature can be determined based on a starting point of the ligature and an ending point of the ligature. Furthermore, the first character can be connected to the second character via a transition. A particular score of the transition can be ascertained based on a difference in length of the first character and the second character. This particular score can be combined with a score corresponding to the particular handwritten characters.

Claims

1. A process for determining a most likely combination of characters, comprising the steps of:

a) obtaining character data which includes information indicative of at least one handwritten character, the character data including at least one set of segmentation points for the at least one handwritten character; and

b) providing a score for each particular character of a set of previously stored characters based on a comparison between the character data and the respective stored particular character,

wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition, and further comprising the steps of;

aa) determining a further score of the transition based on a difference in length of the first character and the second character, wherein step (aa) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character; and

bb) combining the further score with the score corresponding to the particular characters to form a combined score.

2. The process according to claim 1, further comprising the step of:

c) prior to steps (a) and (b), generating the character data after a user completes entering handwritten characters on an input device.

3. The process according to claim 2, wherein the character data corresponds to a handwritten word.

4. The process according to claim 2, wherein the character data is generated when the user disconnects a writing instrument from the input device.

5. The process according to claim 1, wherein the score is provided for each of the previously stored characters by analyzing a predetermined number of the segmentation points of the character data.

6. The process according to claim 1, wherein the character data includes cursive data corresponding to at least one handwritten cursive character and script data corresponding to at least one handwritten script character.

7. The process according to claim 1, wherein the score for each of the previously stored characters is recorded at predetermined time intervals.

8. The process according to claim 1, wherein the score for a current previously stored character is based on previous scores of a word path associated with a current previously stored character.

9. The process according to claim 8, wherein the score is the average of total scores for the previously stored characters corresponding to the word path.

10. The process according to claim 8, further comprising the step of:

d) removing the word path from being considered as the most likely combination of characters, the word path corresponding to a characters sequence.

11. The process according to claim 10, wherein step (d) includes the substep of determining whether the character sequence is to be filtered out based on the score associated with the word path of the character sequence.

12. The process according to claim 11, wherein the determining substep is performed based on a total number of the character sequences.

13. The process according to claim 1, further comprising the step of:

e) normalizing the score, and recording the normalized score in a storage arrangement.

14. The process according to claim 13, wherein step (e) is performed for each of the previously stored characters.

15. The process according to claim 1, wherein the character data corresponds to a sequence of handwritten characters, and further comprising the step of:

f) determining if the sequence exists in a dictionary.

16. The process according to claim 15, wherein step (f) is performed for each of sequences of a plurality of handwritten characters at predetermined time intervals.

17. The process according to claim 15, wherein the dictionary has a tree-type structure, and wherein step (f) is performed by evaluating whether every character of the sequence is provided at a particular node of the tree-type structure of the dictionary.

18. The process according to claim 15, further comprising the step of:

g) if the sequence is not in the dictionary, removing the sequence from being considered as the most likely combination of characters.

19. The process according to claim 15, further comprising the step of:

h) if the sequence is not in the dictionary, arranging the score for the sequence to be at least one of lowered or unchanged.

20. The process according to claim 1, wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a ligature, and further comprising the step of:

i) determining a further score of the ligature based on a starting point of the ligature and an ending point of the ligature.

21. The process according to claim 20, wherein the further score is combined with the score corresponding to the previously stored characters to form a combined score.

22. The process according to claim 1, further comprising the steps of:

j) assigning a vector corresponding to each of the handwritten characters; and

k) comparing each of the vectors to recorded vectors which correspond to the previously stored characters.

23. The process according to claim 22, wherein step (k) is performed by applying a Fisher technique to the vectors corresponding to the handwritten characters.

24. The process according to claim 23, further comprising the step of:

l) comparing the vectors corresponding to the handwritten characters to a cluster of vectors corresponding to the previously stored characters.

25. The process according to claim 24, wherein step (I) is performed by measuring a Euclidean distance from the vector corresponding to the handwritten characters to each of the clusters.

26. The process according to claim 25, wherein the Euclidean distance is determined from a center of each of the clusters.

27. The process according to claim 1, wherein each of the previously stored handwritten characters are assigned to at least one of three classes of characters based on vertical extension dimensions thereof.

28. The process according to claim 27, wherein a first class of the three classes includes first characters which extend beyond a predetermined top position, wherein a second class of the three class includes second characters which extend below a predetermined bottom position, and wherein a third class of the three classes includes third characters which do not extend above the predetermined top position or below the predetermined bottom position.

29. The process according to claim 1, wherein the further score (VBScore) is determined according to the following:

VBScore(<

c₁,c₂>

)=k_H·

M_H(<

c₁,c₂>

)+k_T·

M_T(<

c₁,c₂>

)+k_B·

M_B(<

c₁,c₂>

),

where c₁ is the first character,

c₂ is the second character,

k_H is a height position coefficient,

k_T is a top position coefficient, and

k_B is a bottom position coefficient.

30. The process according to claim 29, wherein the height, top and bottom position coefficients are estimated.

31. The process according to claim 29, wherein the height, top and bottom position coefficients are the same for each pair of the three classes.

32. The process according to claim 29, further comprising the steps of:

m) generating a histogram for each pair of the three classes; and

n) estimating at least one of the height, top and bottom position coefficients based on the histograms.

33. The process according to claim 1, further comprising the step of:

o) prior to steps (a) and (b), segmenting the at least one handwritten character to produce a plurality of segmentation points for the character data.

34. The process according to claim 33, further comprising the step of:

p) adding a further segmentation point between two neighboring segmentation points of the plurality of segmentation points if a Euclidean distance between the two neighboring segmentation points if the Euclidean distance is greater than a predetermined threshold.

35. The process according to claim 33, further comprising the steps of:

q) establishing a segment between two neighboring segmentation points of each pair of the plurality of segmentation points; and

r) assigning a predetermined code to each of the segments using a previously stored list of codes.

36. The process according to claim 35, wherein the predetermined code is assigned to each respective segment based on a similarity of a curvature of the respective segment and a list of previously stored extensions.

37. The process according to claim 36, wherein each respective segment is normalized to be comparable with the previously stored extensions.

38. A process for ascertaining one or more particular characters from character data associated with a handwriting, comprising the steps of:

a) obtaining the character data which includes information indicative of at least one handwritten character; and

b) comparing visual aspects of the at least one handwritten character to visual aspects of each of previously stored characters to ascertain the one or more particular characters

wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition, and further comprising the steps of;

aa) providing a score for each of previously stored characters based on a comparison between the character data and the respective previously stored particular character;

bb) determining a further score of the transition based on a difference in length of the first character and the second character, wherein step (bb) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character; and

cc) combining the further score with the score corresponding to the previously stored characters to form a combined score.

39. The process according to claim 38, wherein the at least one handwritten character includes a plurality of handwritten characters, and wherein step (b) includes comparing at least one of a respective angle and a position of two neighboring characters of the handwritten characters.

40. The process according to claim 38, further comprising the step of:

c) prior to steps (a) and (b), segmenting the at least one handwritten character to produce a plurality of segmentation points for the character data.

41. The process according to claim 40, further comprising the steps of:

d) establishing a segment between two neighboring segmentation points of each pair of the plurality of segmentation points; and

e) assigning a predetermined code to each of the segments using a previously stored list of codes.

42. The process according to claim 41, wherein the predetermined code is assigned to each respective segment based on a similarity of a curvature of the respective segment and a list of previously stored extensions.

43. The process according to claim 38, wherein the score is provided for each of the previously stored characters by analyzing the a predetermined number of the segmentation points of the character data.

44. The process according to claim 43, wherein the character data is generated when the user disconnects a writing instruments from the input device.

45. The process according to claim 38, wherein the score for each previously stored character is recorded at a predetermined time intervals.

46. The process according to claim 38, wherein the score for a current previously stored character is based on previous scores of a word path associated with a current previously stored character.

47. The process according to claim 46, wherein the score is an average of total scores for the previously stored characters corresponding to the word path.

48. The process according to claim 38, wherein the character data includes cursive data corresponding to at least one handwritten cursive character and script data corresponding to at least one handwritten script character.

49. The process according to claim 38, further comprising the steps of:

f) assigning a vector corresponding to each of the handwritten characters; and

g) comparing each of the vectors to recorded vectors which correspond to the previously stored characters.

50. The process according to claim 49, further comprising the step of:

h) comparing the vectors corresponding to the handwritten characters to a cluster of vectors corresponding to the previously stored characters.

51. The process according to claim 50, wherein step (h) is performed by measuring a Euclidean distance from the vector corresponding to the handwritten characters to each of the clusters.

52. The process according to claim 51, wherein the Euclidean distance is determined from a center of each of the clusters.

53. The process according to claim 40, wherein each of the previously stored handwritten characters are assigned to at least one of three classes of characters based on vertical extension dimensions thereof.

54. The process according to claim 53, wherein a first class of the three classes includes first characters which extend beyond a predetermined top position, wherein a second class of the three class includes second characters which extend below a predetermined bottom position, and wherein a third class of the three classes includes third characters which do not extend above the predetermined top position or below the predetermined bottom position.

55. The process according to claim 53, wherein the further score (VBScore) is determined according to the following:

VBScore(<

c₁,c₂>

)=k_H·

M_H(<

c₁,c₂>

)+k_T·

M_T(<

c₁,c₂>

)+k_B·

M_B(<

c₁,c₂>

),

where c₁ is the first character,

c₂ is the second character,

k_H is a height position coefficient,

k_T is a top position coefficient, and

k_B is a bottom position coefficient.

56. The process according to claim 55, wherein the height, top and bottom position coefficients are estimated.

57. The process according to claim 55, wherein the height, top and bottom position coefficients are the same for each pair of the three classes.

58. The process according to claim 57, further comprising the steps of:

i) generating a histogram for each pair of the three classes; and

j) estimating at least one of the height, top and bottom position coefficients based on the histograms.

59. The process according to claim 38, further comprising the step of:

k) prior to steps (a) and (b), segmenting the at least one handwritten character to produce a plurality of segmentation points for the character data.

60. The process according to claim 59, further comprising the step of:

l) adding a further segmentation point between two neighboring segmentation points of the plurality of segmentation points if a Euclidean distance between the two neighboring segmentation points is greater than a predetermined threshold.

61. The process according to claim 59, further comprising the steps of:

m) establishing a segment between two neighboring segmentation points of each pair of the plurality of segmentation points; and

n) assigning a predetermined code to each of the segments using a previously stored list of codes.

62. The process according to claim 61, wherein the predetermined code is assigned to each respective segment based on a similarity of a curvature of the respective segment and a list of previously stored extensions.

63. The process according to claim 62, wherein each respective segment is normalized to be comparable with the previously stored extensions.

64. The process according to claim 38, further comprising the step of:

o) performing a training procedure to obtain at least two variations of the same character.

65. The process according to claim 64, wherein step (of) is performed using a Fisher technique.

66. A process for ascertaining one or more particular characters from character data associated with a handwriting, comprising the steps of:

a) obtaining the character data which includes information indicative of at least one handwritten character; and

b) applying a Fisher Matching procedure on the character data to ascertain the one or more particular characters,

wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition, and further comprising the steps of;

c) providing a score for each of previously stored characters based on a comparison between the character data and the respective previously stored particular character;

d) determining a further score of the transition based on a difference in length of the first character and the second character, wherein step (bb) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character; and

e) combining the further score with the score corresponding to the previously stored characters to form a combined score.

67. A process for ascertaining at least one particular character from character data associated with a handwriting, comprising the steps of:

a) obtaining the character data associated with handwritten characters, a first character of the handwritten characters being connected to a second character of the handwritten characters via a ligature; and

b) determining a first score of the ligature based on a start point of the ligature and an end point of the ligature,

wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition, and further comprising the steps of;

c) providing a second score for each of previously stored characters based on a comparison between the character data and the respective previously stored particular character;

d) determining a third score of the transition based on a difference in length of the first character and the second character, wherein step (bb) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character; and

e) combining the third score with the second score corresponding to the previously stored characters to form a combined score.

68. The process according to claim 67, further comprising the steps of:

c) assigning a vector corresponding to each of the handwritten characters; and

d) comparing each of the vectors to recorded vectors which correspond to previously stored characters.

69. The process according to claim 68, wherein step (d) is performed by applying a Fisher technique to the vectors corresponding to the handwritten characters.

70. The process according to claim 69, further comprising the step of:

e) comparing the vectors corresponding to the handwritten characters to a cluster of vectors corresponding to the previously stored characters.

71. The process according to claim 70, wherein step (e) is performed by measuring a Euclidean distance from the vector corresponding to the handwritten characters to each of the clusters.

72. The process according to claim 71, wherein the Euclidean distance is determined from a center of each of the clusters.

73. A process for ascertaining at least one particular handwritten character from character data associated with handwritten characters, comprising the steps of:

a) obtaining the character data, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition;

b) determining a particular score of the transition based on a difference in length of the first character and the second character, wherein step (b) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character;

c) combining the particular score with a score corresponding to the handwritten characters to form a combined score; and

d) assigning each of previously stored handwritten characters to at least one of three classes of characters based on vertical extension dimensions thereof,

wherein a first class of the three classes includes first characters which extend beyond a predetermined top position, wherein a second class of the three class includes second characters which extend below a predetermined bottom position, and wherein a third class of the three classes includes third characters which do not extend above the predetermined top position or below the predetermined bottom position.

74. A process for ascertaining at least one particular handwritten character from character data associated with handwritten characters, comprising the steps of:

a) obtaining the character data, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition;

b) determining a particular score of the transition based on a difference in length of the first character and the second character, wherein step (b) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character; and

c) combining the particular score with a score corresponding to the handwritten characters to form a combined score,

wherein the particular score (VBScore) is determined according to the following;

VBScore(<

c₁,c₂>

)=k_H·

M_H(<

c₁,c₂>

)+k_T·

M_T(<

c₁,c₂>

)+k_B·

M_B(<

c₁,c₂>

),

where c₁ is the first character,

c₂ is the second character,

k_H is a height position coefficient,

k_T is a top position coefficient, and

k_B is a bottom position coefficient.

75. The process according to claim 74, wherein the height, top and bottom position coefficients are estimated.

76. The process according to claim 74, wherein the height, top and bottom position coefficients are the same for each pair of the three classes.

77. The process according to claim 74, further comprising the steps of:

e) generating a histogram for each pair of the three classes; and

f) estimating at least one of the height, top and bottom position coefficients based on the histograms.

78. A software system which, when executed on a processing device, configures the processing device to determine a most likely combination of characters, the software system comprising:

a processing subsystem which, when executed on the processing device, configures the processing device to perform the following;

a) obtain character data which includes information indicative of at least one handwritten character, the character data including at least one set of segmentation points for the at least one handwritten character,

b) provide a score for each particular character of a set of previously stored characters based on a comparison between the character data and the respective previously stored particular character, wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition,

c) determine a further score of the transition based on a difference in length of the first character and the second character, wherein step (c) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character, and

d) combine the further score with the score corresponding to the particular characters to form a combined score.

79. The software system according to claim 78, wherein, when executed on the processing device, the processing subsystem is capable of generating the character data after a user completes entering handwritten characters on an input device.

80. The software system according to claim 79, wherein the character data corresponds to a handwritten word.

81. The software system according to claim 79, wherein the score is provided for each of the previously stored characters by analyzing a predetermined number of the segmentation points of the character data.

82. The software system according to claim 79, wherein the character data is generated when the user disconnects a writing instrument from the input device.

83. The software system according to claim 78, wherein the character data includes cursive data corresponding to at least one handwritten cursive character and script data corresponding to at least one handwritten script character.

84. The software system according to claim 78, wherein the score for each of the previously stored characters is recorded at predetermined time intervals.

85. The software system according to claim 78, wherein the score for a current previously stored character is based on previous scores of a word path associated with a current previously stored character.

86. The software system according to claim 85, wherein the score is the average of total scores for the previously stored characters corresponding to the word path.

87. The software system according to claim 85, wherein, when in operation, the processing subsystem is configured to remove the word path from being considered as the most likely combination of characters, the word path corresponding to a characters sequence.

88. The software system according to claim 87, wherein, when in operation, the processing subsystem is configured to determines whether the character sequence is to be filtered out based on the score associated with the word path of the character sequence.

89. The software system according to claim 88, wherein, when in operation, the processing subsystem is configured to make the determination regarding the character sequences based on a total number of the character sequences.

90. The software system according to claim 78, wherein, when in operation, the processing subsystem is configured to normalize the score, and record the normalized score in a storage arrangement.

91. The software system according to claim 78, wherein the character data corresponds to a sequence of handwritten characters, and wherein, when in operation, the processing subsystem is configured to determine if the sequence exists in a dictionary.

92. The software system according to claim 91, wherein the determination regarding the existence of the sequence is performed for each of sequences of a plurality of handwritten characters at predetermined time intervals.

93. The software system according to claim 91, wherein the dictionary has a tree-type structure, and wherein the determination regarding the existence of the sequence is performed by evaluating whether every character of the sequence is provided at a particular node of the tree-type structure of the dictionary.

94. The software system according to claim 91, wherein, when in operation, the processing subsystem is configured to remove the sequence from being considered as the most likely combination of characters if the sequence is not in the dictionary.

95. The software system according to claim 91, wherein, when in operation, the processing subsystem is configured to arrange the score for the sequence to be at least one of lowered or unchanged if the sequence is not in the dictionary.

96. The software system according to claim 95, wherein the further score is combined with the score corresponding to the previously stored characters to form a combined score.

97. The software system according to claim 78, wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a ligature, and wherein, when in operation, the processing subsystem is configured to determine the further score of the ligature based on a starting point of the ligature and an ending point of the ligature.

98. The software system according to claim 78, wherein, when in operation, the processing subsystem is configured to:

e) assign a vector corresponding to each of handwritten characters, and

f) compare each of the vectors to recorded vectors which correspond to the previously stored characters.

99. The software system according to claim 98, wherein the comparison is performed by applying a Fisher technique to the vectors corresponding to the handwritten characters.

100. The software system according to claim 99, wherein, when in operation, the processing subsystem is configured to compare the vectors corresponding to the handwritten characters to a cluster of vectors corresponding to the previously stored characters.

101. The software system according to claim 100, wherein the comparison of the vectors is performed by measuring a Euclidean distance from the vector corresponding to the handwritten characters to each of the clusters.

102. The software system according to claim 101, wherein the Euclidean distance is determined from a center of each of the clusters.

103. The software system according to claim 78, wherein each of the previously stored handwritten characters are assigned to at least one of three classes of characters based on vertical extension dimensions thereof.

104. The software system according to claim 103, wherein a first class of the three classes includes first characters which extend beyond a predetermined top position, wherein a second class of the three class includes second characters which extend below a predetermined bottom position, and wherein a third class of the three classes includes third characters which do not extend above the predetermined top position or below the predetermined bottom position.

105. The software system according to claim 78, wherein the further score (VBScore) is determined according to the following:

VBScore(<

c₁,c₂>

)=k_H·

M_H(<

c₁,c₂>

)+k_T·

M_T(<

c₁,c₂>

)+k_B·

M_B(<

c₁,c₂>

),

where c₁ is the first character,

c₂ is the second character,

k_H is a height position coefficient,

k_T is a top position coefficient, and

k_B is a bottom position coefficient.

106. The software system according to claim 105, wherein the height, top and bottom position coefficients are estimated.

107. The software system according to claim 105, wherein the height, top and bottom position coefficients are the same for each pair of the three classes.

108. The software system according to claim 105, wherein, when in operation, the processing subsystem is configured to:

g) generate a histogram for each pair of the three classes, and

h) estimate at least one of the height, top and bottom position coefficients based on the histograms.

109. The software system according to claim 78, wherein, when in operation, the processing subsystem is configured to segment the at least one particular character to produce a plurality of segmentation points for the character data.

110. The software system according to claim 109, wherein, when in operation, the processing subsystem is configured to add a further segmentation point between two neighboring segmentation points of the plurality of segmentation points if a Euclidean distance between the two neighboring segmentation points is greater than a predetermined threshold.

111. The software system according to claim 109, wherein, when in operation, the processing subsystem is configured to:

j) establish a segment between two neighboring segmentation points of each pair of the plurality of segmentation points and

k) assign a predetermined code to each of the segments using a previously stored list of codes.

112. The software system according to claim 111, wherein the predetermined code is assigned to each respective segment based on a similarity of a curvature of the respective segment and a list of previously stored extensions.

113. The software system according to claim 112, wherein each respective segment is normalized to be comparable with the previously stored extensions.

114. A software system which, when executed on a processing device, configures the processing device to ascertain one or more particular characters from character data associated with a handwriting, the software system comprising:

a processing subsystem which, when executed on the processing device, configures the processing device to perform the following steps;

a) obtain the character data which includes information indicative of at least one handwritten character,

b) compare visual aspects of the at least one handwritten character to visual aspects of each of previously stored characters to ascertain the one or more particular characters, wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition,

c) provide a score for each of previously stored characters based on a comparison between the character data and the respective previously stored particular character,

d) determine a further score of the transition based on a difference in length of the first character and the second character, wherein step (bb) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character, and

e) combine the further score with the score corresponding to the previously stored characters to form a combined score.

115. A software system which, when executed on a processing device, configures the processing device to ascertain one or more particular characters from character data associated with a handwriting, the software system comprising:

a processing subsystem which, when executed on the processing device, configures the processing device to perform the following steps;

a) obtain the character data which includes information indicative of at least one handwritten character,

b) apply a Fisher Matching procedure on the character data to ascertain the one or more particular characters, wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition,

c) provide a score for each of previously stored characters based on a comparison between the character data and the respective previously stored particular character,

d) determine a further score of the transition based on a difference in length of the first character and the second character, wherein step (bb) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character, and

e) combining the further score with the score corresponding to the previously stored characters to form a combined score.

116. A software system which, when executed on a processing device, configures the processing device to ascertain at least one particular character from character data associated with handwritten characters, the software system comprising:

a processing subsystem which, when executed on the processing device, configures the processing device to perform the following steps;

a) obtain the character data associated with handwritten characters, a first character of the handwritten characters being connected to a second character of the handwritten characters via a ligature,

b) determine a first score of the ligature based on a starting point of the ligature and an ending point of the ligature, wherein the at least one handwritten character includes a plurality of handwritten characters, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition,

c) provide a second score for each of previously stored characters based on a comparison between the character data and the respective previously stored particular character,

d) determine a third score of the transition based on a difference in length of the first character and the second character, wherein step (bb) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character; and

e) combine the third score with the second score corresponding to the previously stored characters to form a combined score.

117. A software system which, when executed on a processing device, configures the processing device to ascertain at least one particular handwritten character from character data associated with handwritten characters, the software system comprising:

a processing subsystem which, when executed on the processing device, configures the processing device to perform the following steps;

a) obtain the character data, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition,

b) determine a particular score of the transition based on a difference in length of the first character and the second character, wherein step (b) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character,

c) combine the particular score with a score corresponding to the particular handwritten characters to form a combined score; and

d) assigning each of previously stored handwritten characters to at least one of three classes of characters based on vertical extension dimensions thereof,

wherein a first class of the three classes includes first characters which extend beyond a predetermined top position, wherein a second class of the three class includes second characters which extend below a predetermined bottom position, and wherein a third class of the three classes includes third characters which do not extend above the predetermined top position or below the predetermined bottom position.

118. A software system which, when executed on a processing device, configures the processing device to ascertain at least one particular handwritten character from character data associated with handwritten characters, the software system comprising:

a processing subsystem which, when executed on the processing device, configures the processing device to perform the following steps;

a) obtain the character data, wherein a first character of the handwritten characters is connected to a second character of the handwritten characters via a transition,

b) determine a particular score of the transition based on a difference in length of the first character and the second character, wherein step (b) is determined based on a height difference ratio (M_H), a top difference ratio (M_T) and a bottom difference ratio (M_B) of the first character and the second character, and

c) combine the particular score with a score corresponding to the particular handwritten characters to form a combined score,

wherein the particular score (VBScore) is determined according to the following;

VBScore(<

c₁,c₂>

)=k_H·

M_H(<

c₁,c₂>

)+k_T·

M_T(<

c₁,c₂>

)+k_B·

M_B(<

c₁,c₂>

),

where c₁ is the first character,

c₂ is the second character,

k_H is a height position coefficient,

k_T is a top position coefficient, and

k_B is a bottom position coefficient.