Preprocessing of multi-line rotated electronic ink

• US 7,412,095 B2
• Filed: 10/24/2005
• Issued: 08/12/2008
• Est. Priority Date: 05/10/2002
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Hullender, Gregory N., Abdulkader, Ahmad
• Primary Examiner(s): Mehta; Bhavesh
• Assistant Examiner(s): Strege; John B
• Application Number:
  US11/257,332
• Publication Number:
  US 20060034516A1
• Time in Patent Office: 1,023 Days
• Field of Search: 382/177, 382186-189, 382294-296
• US Class Current: 382/186
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/1423 the instrument generating sequences of position coordinates corresponding to handwriting preprocessing or recognising digital ink G06V30/32
• CPC Class Codes: G06V 30/1478 of characters or characters lines

Abstract

An electronic ink object is preprocessed for improving handwriting recognition results. An electronic ink object, possibly having multiple lines of non-horizontal ink, is preprocessed by segmenting or separating the electronic ink object into one or more lines, computing a writing angle of each line, and rotating each line so that it is horizontal.

Claims

1. A method of preprocessing an electronic ink object comprising the steps of:

segmenting the electronic ink object into a plurality of lines;

estimating a writing angle of each line; and

rotating each line to make horizontal;

and creating a model of each line for interpretation and further evaluation between the points of local extrema of the segmented electronic ink object and the model to reduce incorrect instances of recognition;

wherein the model includes;

employing an electronic ink feature generator;

employing a line breaking classifier;

employing a mapping database; and

an ink transformer.

2. The method of preprocessing an electronic ink object of claim 1 further comprising the step of:

recognizing the segmented and rotated electronic ink object to identify at least one character.

3. The method of preprocessing an electronic ink object of claim 1 wherein the electronic ink object comprises at least one ink stroke and the step of segmenting includes the step of (i) identifying a plurality of features for each ink stroke of the electronic ink object;

and (ii) classifying each ink stroke as part of a new line, a current line, or a previous line.

4. The method of preprocessing an electronic ink object of claim 3 wherein the step of classifying is performed using a neural network.

5. The method of preprocessing an electronic ink object of claim 3 wherein the step of classifying is performed using a hidden Markov model.

6. The method of preprocessing an electronic ink object of claim 3 wherein the step of classifying is performed using a decision tree.

7. The method of preprocessing an electronic ink object of claim 3 wherein the step of classifying is performed using a Bayesian net.

8. The method of preprocessing an electronic ink object of claim 3 wherein the step of classifying is performed using a K Nearest Neighbor (KNN).

9. The method of preprocessing an electronic ink object of claim 3 wherein the step of classifying is performed using an Expert System.

10. The method of preprocessing an electronic ink object of claim 3 wherein if the ink stroke is part of the new line, the step of estimating includes the step of computing a slope of the new line.

11. The method of preprocessing an electronic ink object of claim 3 wherein if the ink stroke is part of the new line, the step of estimating includes the step of computing a slope of a centerline of the new line using a Least Mean Squares algorithm.

12. The method of preprocessing an electronic ink object of claim 1 wherein the step of rotating is performed using a rotation matrix.

13. A computer-readable medium having computer-executable instructions for performing the steps recited in claim 1.

22. The method of preprocessing an electronic ink object of claim 1 wherein the model of each line for interpretation is a Gaussian model.

23. The method of preprocessing an electronic ink object of claim 1 wherein the segmenting the electronic ink object into a plurality of lines based upon at least one occurrence of a cusp within the at least one ink stroke, and defined as differences between tangents on successive points of a chord of the at least one ink stroke.

14. A system for processing an electronic ink object having at least one ink stroke comprising in combination:

an input that receives the electronic ink object;

an electronic ink feature generator that receives the inputted electronic ink object and generates at least one feature for the at least one ink stroke of the electronic ink object;

a classifier that classifies the at least one ink stroke of the electronic ink object into a plurality of lines based on the feature of the ink stroke; and

an ink transformer that rotates each of the plurality of lines of electronic ink corresponding to a center of the ink object from a non-horizontal to a horizontal relationship according to each of the plurality of lines having a spatial orientation deviating from a horizontal placement upon a graphic display and better facilitate preprocessed electronic ink object to written character recognition.

15. The system of claim 14 wherein the classifier is a neural network.

16. The system of claim 14 wherein the classifier is a hidden Markov model.

17. The system of claim 14 wherein the classifier is a decision tree.

18. The system of claim 14 wherein the classifier is a Bayesian net.

19. The system of claim 14 wherein the classifier is a K Nearest Neighbor (KNN).

20. The system of claim 14 wherein the classifier is an Expert System.

21. The system of claim 14 further comprising:

a user interface that generates the electronic ink object.

24. The system of claim 14 wherein the classifier that classifies the at least one ink stroke of the electronic ink object into a plurality of lines based upon at least one occurrence of a cusp within the at least one ink stroke, and defined as differences between tangents on successive points of a chord of the at least one ink stroke.