Statistical Online Character Recognition

• US 20100329562A1
• Filed: 06/30/2009
• Published: 12/30/2010
• Est. Priority Date: 06/30/2009
• Status: Active Grant

• Current Assignee: Oracle America, Inc. (Oracle Corporation)
• Original Assignee: Oracle America, Inc. (Oracle Corporation)
• Inventors: Zhu, Feng Drake
• Granted Patent: US 8,391,613 B2
• Time in Patent Office: Days
• US Class Current: 382/187
• CPC Class Codes: G06V 30/347 Sampling; Contour coding; Stroke extraction
• CPC Class Codes: G06V 30/36 Matching; Classification

Abstract

A statistical system and method for generating patterns and performing online handwriting recognition based on those patterns. A plurality of predetermined patterns may be generated by performing feature extraction operations on one or more character samples utilizing a Gabor filter. An online handwritten character may be acquired. The online handwritten character may be pre-processed. One or more feature extraction operations, utilizing a Gabor filter, may be performed on the online handwritten character to produce a feature vector. One or more patterns may be generated, using a statistical algorithm, for the online handwritten character, based on the feature vector. The online handwritten character may be statistically classified based on a comparison between the one or more patterns generated for the online handwritten character and the plurality of predetermined patterns.

Claims

1. A method for generating patterns for use in online character recognition, the method comprising:

performing one or more pre-processing operations on a character sample;

performing one or more feature extraction operations on the character sample, wherein the one or more feature extraction operations produce a feature vector for the character sample, wherein said performing the one or more feature extraction operations utilizes a Gabor filter;

performing statistical training to generate patterns based on the feature vector of the character sample;

storing the patterns in a memory, wherein the patterns are configured to be used to recognize handwritten characters.

2. The method of claim 1,

wherein the one or more feature extraction operations comprise filtering directional pattern images using a Gabor filter, wherein the filtered directional pattern images are used to form the feature vector.

3. The method of claim 1,

wherein the one or more feature extraction operations comprise:

extracting directional features based on the direction of the character sample'"'"'s points;

generating directional pattern images based on the directional features;

filtering the directional pattern images using a Gabor filter; and

forming the feature vector based on the filtered directional pattern images.

4. The method of claim 1,

wherein the one or more pre-processing operations comprise adding imaginary strokes and adding imaginary points.

5. The method of claim 1,

wherein the one or more pre-processing operations comprise three or more of:

noise point removal, linear size normalization, nonlinear shape normalization, adding imaginary strokes, and adding imaginary points.

6. The method of claim 1,

wherein the one or more pre-processing operations comprise noise point removal, linear size normalization, nonlinear shape normalization, adding imaginary strokes, adding imaginary points, re-sampling, and smoothing.

7. The method of claim 1,

wherein the statistical training comprises a k-means clustering algorithm.

8. The method of claim 1,

wherein the statistical training comprises a clustering algorithm.

9. The method of claim 8,

wherein the feature vector comprises a plurality of directional vectors, wherein at least two of the directional vectors are weighted differently.

10. A memory medium comprising program instructions for generating patterns for use in online character recognition, wherein the program instructions are executable to:

perform one or more pre-processing operations on a character sample;

perform one or more feature extraction operations on the character sample, wherein the one or more feature extraction operations produce a feature vector for the character sample, wherein said performing the one or more feature extraction operations utilizes a Gabor filter;

perform statistical training to generate patterns based on the feature vector of the character sample;

store the patterns in a memory, wherein the patterns are configured to be used to recognize handwritten characters.

11. The memory medium of claim 10,

wherein in performing the one or more feature extraction operations the program instructions are executable to filter directional pattern images using a Gabor filter, wherein the filtered directional pattern images are used to form the feature vector.

12. The memory medium of claim 10,

wherein in performing the one or more feature extraction operations, the program instructions are executable to:

extract directional features based on the direction of the character sample'"'"'s points;

generate directional pattern images based on the directional features;

filter the directional pattern images using a Gabor filter; and

form the feature vector based on the filtered directional pattern images.

13. The memory medium of claim 10,

wherein the one or more pre-processing operations comprise adding imaginary strokes and adding imaginary points.

14. The memory medium of claim 10,

wherein the one or more pre-processing operations comprise three or more of:

noise point removal, linear size normalization, nonlinear shape normalization, adding imaginary strokes, and adding imaginary points.

15. A method for recognizing handwritten characters using predetermined patterns, comprising:

generating the plurality of predetermined patterns, wherein the plurality of predetermined patterns are generated by performing one or more feature extraction operations on the character sample utilizing a Gabor filter, wherein the plurality of predetermined patterns are stored in a memory;

acquiring an online handwritten character;

pre-processing the online handwritten character;

extracting features of the online handwritten character, thereby determining a feature vector;

generating, using a statistical algorithm, one or more patterns for the online handwritten character based on the feature vector;

classifying the online handwritten character, comprising determining a character corresponding to the online handwritten character based on the generated one or more patterns and the plurality of predetermined patterns.

16. The method of claim 15,

wherein the plurality of predetermined patterns are generated based on a statistical character recognition method.

17. The method of claim 15, wherein said generating the plurality of predetermined patterns comprises:

performing one or more pre-processing operations on a character sample;

performing one or more feature extraction operations on the character sample, wherein the one or more feature extraction operations produce a feature vector for the character sample, wherein said performing the one or more feature extraction operations utilizes a Gabor filter;

performing statistical training to generate patterns based on the feature vector of the character sample;

storing the plurality of predetermined patterns in the memory, wherein the plurality of predetermined patterns are configured to be used to recognize handwritten characters.

18. The method of claim 15,

wherein said acquiring the online handwritten character comprises receiving input data collected from at least one input stroke in a pen-input device.

19. The method of claim 15,

wherein the online handwritten character comprises one or more strokes.

20. The method of claim 15,

wherein the online handwritten character includes information about an order in which the points of the online handwritten character were input.