Patent US4024500A successfully refreshed from Google Patents.
← Back to Patent Explorer
US4024500A Expired / Lapsed US ⏱️ 0.0 years remaining

Segmentation mechanism for cursive script character recognition systems

Assignee: International Business Machines Corp
•
Filing: 1975-12-31
•
Granted: 1977-05-17
•
Expires: 1995-12-31
📄 Original PDF 🌐 Google Patents 🔄 Sync Now

Abstract

Abstract A method and apparatus for effecting character segmentation in a cursive script handwriting analysis system which comprises obtaining the continuous x and y coordinates and the x and y velocities of a writing instrument forming said cursive character. Continuously averaging the x displacement associated with all the x and y extremal points of a handwritten character where the x or the y velocity equals zero. Successively examining the x displacement for each x-extremal and determining if the x displacement of said extremal exceeds a predetermined threshold value relative to the average x displacement of the current character and if so indicating that said new extremal is located in the next character and that a segmentation mark should be placed at a predetermined distance along the ligature between the just analyzed extremal and the previously analyzed extremal. The results of this segmentation are then communicated to a character recognition mechanism whose efficiency and accuracy is greatly enhanced by said segmentation indication. The system as designed is also able to follow deviations from the baseline and mid-zone line as the writing progresses.

Inventors

  • 👤 Noel Martin Herbst
  • 👤 John Henry Morrissey

Bibliographic Details

Application Number —
Jurisdiction US
Data Source google_patents
Last Synchronized 2026-10-27 10:57
CPC Classifications
G; G06; G06V; G06V30/00; G06V30/10; G06V30/22; G06V30/226; G06V30/2268; G06V30/14; G06V30/148; G06V30/15

Claims Analysis

Total Claims: 21
Claim 1
Independent Claim 135 words
A character segmentation mechanism for cursive handwriting which comprises: means for obtaining continuous X and Y coordinate data and the instantaneous X and Y velocities associated therewith as a writing stylus is moving, means for detecting when a Y extremal is encountered (Y velocity = 0), means for detecting when an X extremal (X velocity = 0) is encountered, means for successively averaging the X displacement values associated with a particular segment whenever either extremal is encountered, whereby the average is updated each time, means for comparing the X displacement value associated with each X extremal to see if its magnitude is more than a predetermined threshold value above the current average X displacement value, and means for specifying that a segmentation should occur when the said displacement value is greater than said predetermined amount.
Claim 2
Dependent Claim 135 words
A character segmentation mechanism for cursive handwriting which comprises: means for obtaining continuous X and Y coordinate data and the instantaneous X and Y velocities associated therewith as a writing stylus is moving, means for detecting when a Y extremal is encountered (Y velocity = 0), means for detecting when an X extremal (X velocity = 0) is encountered, means for successively averaging the X displacement values associated with a particular segment whenever either extremal is encountered, whereby the average is updated each time, means for comparing the X displacement value associated with each X extremal to see if its magnitude is more than a predetermined threshold value above the current average X displacement value, and means for specifying that a segmentation should occur when the said displacement value is greater than said predetermined amount.
Claim 3
Dependent Claim 68 words
A character segmentation mechanism as set forth in claim 1 including means for computing a segmentation point which comprises: giving a first predetermined weighting to the X displacement of the previously encountered X extremal value and a second predetermined weighting to the X displacement for the extremal value when a segmentation decision was made, and adding said weight displacement values to obtain the X displacement of the segmentation.
Claim 4
Dependent Claim 95 words
A character segmentation mechanism as set forth in claim 1 including means for continuously examining the Y displacements of all sampled displacement values for a given cursive handwriting sample, and determining if the Y displacement is either less than or greater than a predetermined vertical zone value, means for storing the X displacement value associated with said Y displacement value, when said determination is affirmative, means for using said X displacement value so stored for X displacement averaging operations for all extremals whose Y displacement values are greater or less than said predetermined zone values.
Claim 5
Dependent Claim 75 words
A character segmentation mechanism as set forth in claim 3 including means for continuously updating said predetermined vertical Zone values wherein said predetermined zone values represent the Y displacements for the top and bottom, respectively, of lower case script characters not having ascenders or descenders, which step includes averaging the respective Y displacement, for any Y extremal found to be within a predetermined distance from said previously predetermined displacements, with said previous predetermined zone value.
Claim 6
Dependent Claim 32 words
A character segmentation mechanism as set forth in claim 4 including means for adding a predetermined threshold value to said predetermined Y displacement values prior to making said greater than/less than comparison.
Claim 7
Dependent Claim 109 words
A character segmentation method as set forth in claim 1 including means for successively evaluating the Y displacement values of all sampled X-Y displacement values for the corresponding sampling points for a given cursive handwriting sample, and determining if the Y displacement is either less than or greater than a predetermined vertical zone value having an upper and a lower boundary value, storing the X displacement value associated with said Y displacement value, when said determination is affirmative, using said X displacement value so stored for X displacement averaging operations for all X and Y extremals whose Y displacement values are greater or less than said predetermined zone values.
Claim 8
Dependent Claim 77 words
A character segmentation method as set forth in claim 6 including the step of continuously updating said predetermined vertical zone values wherein said predetermined zone values represent the Y displacements for the top and bottom, respectively, of lower case script characters not having ascenders or descenders, said step further including averaging the respective Y displacement, for any Y extremal found to be within a predetermined distance from said previously predetermined displacements with said previous predetermined zone value.
Claim 9
Dependent Claim 33 words
A character segmentation method as set forth in claim 7 including the steps of combining a predetermined threshold value with said predetermined Y displacement values prior to making said greater than/less than comparison.
Claim 10
Dependent Claim 110 words
A character segmentation mechanism for cursive handwriting including: writing stylus, tablet and data generation means for providing continuous samples of X and Y displacements of said moving stylus and X and Y velocity data associated with each displacement point, means for detecting the X extremals (x = 0) and the Y extremals (y = 0), means utilizing the X coordinates of both the X and Y extremals for determining an average character midpoint on the X axis, and, means for examining the X coordinate of each X extremal for determining if such extremal lies within a predetermined X boundary for a given character based on said average character midpoint determination.
Claim 11
Dependent Claim 110 words
A character segmentation mechanism for cursive handwriting including: writing stylus, tablet and data generation means for providing continuous samples of X and Y displacements of said moving stylus and X and Y velocity data associated with each displacement point, means for detecting the X extremals (x = 0) and the Y extremals (y = 0), means utilizing the X coordinates of both the X and Y extremals for determining an average character midpoint on the X axis, and, means for examining the X coordinate of each X extremal for determining if such extremal lies within a predetermined X boundary for a given character based on said average character midpoint determination.
Claim 12
Dependent Claim 63 words
A character segmentation mechanism as set forth in claim 9 wherein the means for determining said average character midpoint comprises means for successively averaging the X displacement value for every extremal encountered with an initial X displacement value whereby as more extremal points for a given character are encountered, the average midpoint becomes more truly representative of the mid location of the character.
Claim 13
Dependent Claim 58 words
A character segmentation mechanism as set forth in claim 10 including means for comparing the X displacement value associated with each X extremal encountered with the current character average midpoint value to determine if it is within a predetermined distance thereof, and means for indicating to the system that a segmentation should occur when such condition is encountered.
Claim 14
Dependent Claim 76 words
A character segmentation mechanism as set forth in claim 11 including means for determining where the segmentation is to occur between two characters after a decision to segment has been made, said determining means including means for weighting and adding the X displacement of the last encountered X extremal to a weighted value for the X displacement of the currently encountered X extremal which it has been found lies outside of a predetermined character width boundary.
Claim 15
Dependent Claim 31 words
A character segmentation mechanism as set forth in claim 12 including means for determining that a predetermined number of extremals has been encountered before a decision to segment can be made.
Claim 16
Dependent Claim 45 words
A character segmentation mechanism as set forth in claim 13 including means for considering the X displacements of those extremals belonging to ascender or descender strokes to be the values where they cross the midzone and base line, respectively, for the handwriting sample being analyzed.
Claim 17
Dependent Claim 82 words
A character segmentation mechanism as set forth in claim 14 including means for examining the Y displacement value associated with each Y extremal and means for determining if said Y displacement value is within a threshold range of an average midzone value or within a threshold range of an average base line value, and means responsive to such a determination for modifying the average midzone or base line values, respectively, in accordance with the Y displacement of the just encountered Y extremal.
Claim 18
Dependent Claim 65 words
A segmentation mechanism as set forth in claim 15 including means for specifying that the initial average X displacement value specifying an estimated midpoint for each character is either the beginning X displacement where the character is the first character in the string of connected characters, or the X displacement of the segmentation point where it is not the first character in such a string.
Claim 19
Independent Claim 133 words
A method for effecting character segmentation in a cursive script handwriting analysis system which comprises: obtaining continuous X and Y coordinates and the instantaneous X and Y velocities related thereto as a writing stylus is moving, detecting when a Y extremal is encountered (Y velocity = 0), detecting when a X extremal (X velocity = 0) is encountered, successively averaging the X displacement values associated with a particular segment whenever either extremal is encountered, whereby said X displacement value is successively updated, comparing the X displacement value associated with each X extremal to see if it is more than a predetermined magnitude greater than the current average X displacement value for the segment being evaluated, and deciding that a segmentation should occur when the said displacement magnitude is greater than said predetermined amount.
Claim 20
Independent Claim 133 words
A method for effecting character segmentation in a cursive script handwriting analysis system which comprises: obtaining continuous X and Y coordinates and the instantaneous X and Y velocities related thereto as a writing stylus is moving, detecting when a Y extremal is encountered (Y velocity = 0), detecting when a X extremal (X velocity = 0) is encountered, successively averaging the X displacement values associated with a particular segment whenever either extremal is encountered, whereby said X displacement value is successively updated, comparing the X displacement value associated with each X extremal to see if it is more than a predetermined magnitude greater than the current average X displacement value for the segment being evaluated, and deciding that a segmentation should occur when the said displacement magnitude is greater than said predetermined amount.
Claim 21
Dependent Claim 68 words
A character segmentation method as set forth in claim 17 including the steps of computing a segmentation point which comprises giving a first predetermined weighting to the X displacement of the previously encountered extremal value and a second predetermined weighting to the X displacement for the external value when a segmentation decision was made, and adding said weighted displacement values to obtain the X displacement of the segmentation.

References Cited (Backward Citations)

US3111646A Cited
US3344399A Cited

USPTO Annuity & Maintenance Windows

Annuity Stage Window Opens Payment Deadline Surcharge Period Until Estimated Fee Status
3.5 Years (Year 4) 1980-05-16 1980-11-15 1981-05-17 $2000.00 Window Closed / Lapsed
7.5 Years (Year 8) 1984-05-16 1984-11-15 1985-05-17 $3760.00 Window Closed / Lapsed
11.5 Years (Year 12) 1988-05-16 1988-11-15 1989-05-17 $7700.00 Window Closed / Lapsed

Assigned Categories & Tags

No categories assigned to this patent yet.

Assign Category Tag