Method and apparatus for arc segmentation in handwriting recognition

• US 5,610,996 A
• Filed: 07/13/1994
• Issued: 03/11/1997
• Est. Priority Date: 04/15/1991
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Eller, Marlin J.
• Primary Examiner(s): Couso, Jose L.
• Application Number:
  US08/274,354
• Time in Patent Office: 972 Days
• Field of Search: 382/3, 382/9, 382/13, 382/48, 382/119, 382/173, 382/187, 382/282, 178/18-20
• US Class Current: 382/187
• CPC Class Codes: G06V 30/347 Sampling; Contour coding; Stroke extraction
• CPC Class Codes: G06V 30/373 using a special pattern or subpattern alphabet

Abstract

According to principles of the invention, a method and apparatus are provided that organizes sample text into a format that facilitates character recognition. Sample text is analyzed for those features that define characters. Features that represent useful data for character recognition are stored and analyzed but features not useful for character recognition can be discarded. The sample text is segmented into arcs at each Y-extrema. A center of gravity is calculated for each arc. The arc features are saved as descriptive of the sample text for use in character recognition.

Claims

1. A computer-implemented method of recognizing a handwritten character comprising the steps of:

inputting into a computer a plurality of data points representing the handwritten character, the data points corresponding to X-Y coordinate points in a path traveled by a handwriting instrument as the character is written, the handwritten character including a first arc having first and second end points;

detecting a change in direction of curvature of the first arc from a clockwise direction to a counterclockwise direction, the change of direction of curvature indicating an inflection point, wherein the detecting step includes;

selecting a first vector heading into a first data point of the plurality of data points of the first arc;

selecting a second vector heading out of the first data point and into a second data point of the plurality of data points of the first arc;

selecting a third vector heading out of the second data point;

computing a first angle between the first and second vectors, the first angle having a positive or negative sign depending on whether the first angle is clockwise or counterclockwise;

computing a second angle between the second and third vectors, the second angle having a positive or negative sign depending on whether the second angle is clockwise or counterclockwise;

determining whether the second angle exceeds a predetermined turning point threshold value;

if the second angle exceeds the turning point threshold value, then segmenting the first arc into second and third arcs at the second data point;

if the second angle does not exceed the turning point threshold value, then determining whether the second angle is opposite in sign to the first angle; and

if the second angle is opposite in sign to the first angle, then segmenting the arcs into the second and third arcs at the second data point, the second data point being the inflection point;

segmenting the first arc into second and third arcs at the inflection point, the inflection point becoming an end point in each of the second and third arcs;

determining a representative set of coordinate points for a selected one of the arcs of the handwritten character, the representative set having fewer coordinate points than the selected arc;

comparing the representative set to arcs in a plurality of reference characters; and

outputting a representation of a reference character based on a result obtained in the comparing step.

2. The method according to claim 1, further comprising de-hooking and smoothing said data points by:

determining a difference in distance between a point cluster location and a sequentially following data point;

selecting the sequentially following data point as a new point cluster if it is spaced more than a threshold distance from the point cluster location;

averaging the data point'"'"'s location into the point cluster location if it is spaced less than the threshold distance from the point cluster location; and

treating the point cluster locations as individual points in the handwritten character for segmenting the handwritten character into arcs.

3. The method according to claim 1, further comprising:

calculating a flatness value indicative of the flatness of the second and third arcs; and

combining the second and third arcs into the first arc if the flatness value calculated is less than a threshold flatness value.

4. The method according to claim 3 wherein the step of calculating a flatness value includes calculating a cumulative curvature value for the second arc and a cumulative curvature value for the third arc and calculating a difference between the cumulative curvature values.