License plate optical character recognition method and system

• US 8,644,561 B2
• Filed: 01/18/2012
• Issued: 02/04/2014
• Est. Priority Date: 01/18/2012
• Status: Active Grant

• Current Assignee: Conduent Business Services, LLC (Conduent, Inc.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Burry, Aaron Michael, Kozitsky, Vladimir, Paul, Peter
• Primary Examiner(s): Fitzpatrick, Atiba O
• Application Number:
  US13/352,554
• Publication Number:
  US 20130182910A1
• Time in Patent Office: 748 Days
• Field of Search: 382100-107, 348/113, 348169-171
• US Class Current: 382/107
• CPC Class Codes: G06F 18/243 relating to the number of classes
• CPC Class Codes: G06V 20/625 License plates
• CPC Class Codes: G06V 30/10 Character recognition

Abstract

A method and system for recognizing a license plate character utilizing a machine learning classifier. A license plate image with respect to a vehicle can be captured by an image capturing unit and the license plate image can be segmented into license plate character images. The character image can be preprocessed to remove a local background variation in the image and to define a local feature utilizing a quantization transformation. A classification margin for each character image can be identified utilizing a set of machine learning classifiers each binary in nature, for the character image. Each binary classifier can be trained utilizing a character sample as a positive class and all other characters as well as non-character images as a negative class. The character type associated with the classifier with a largest classification margin can be determined and the OCR result can be declared.

Claims

1. A method for recognizing a license plate character, said method comprising:

capturing a license plate image with respect to a vehicle utilizing an image capturing unit and thereafter segmenting said license plate image into a license plate character image;

pre-processing said license plate character image to remove a local background variation with respect to said license plate character image and to define a local feature thereof utilizing a quantization transformation; and

identifying a classification margin for each character image utilizing a set of machine learning classifiers each binary in nature for said character image in order to thereafter determine a character type associated with at least one machine learning classifier among said set of machine learning classifiers with a largest classification margin to declare an optical character recognition result.

2. The method of claim 1 further comprising utilizing a multi-class machine learning classifier instead of said set of machine learning classifiers in order to declare said optical character recognition result.

3. The method of claim 1 further comprising training each binary classifier utilizing a character sample as a positive class and other characters as well as non-character images as a negative class.

4. The method of claim 1 further comprising:

grouping said license plate character image into an image training set of each character type; and

performing training by applying said quantization transformation and said set of machine learning classifiers for each training set and each character image.

5. The method of claim 1 wherein said quantization transformation comprises a successive mean quantization transformation.

6. The method of claim 1 wherein said set of machine learning classifiers comprises a set of sparse network of winnows machine based image classifiers.

8. The method of claim 1 wherein defining said local feature utilizing said quantization transformation, further comprises:

defining a local window of operation having a plurality of local image regions and determining a mean value in said plurality of local image regions;

setting a most significant bit of each pixel in an output region to 1 wherein a gray level with respect to an input region is larger than a mean value; and

setting said most significant bit of each pixel in said output region to 0 wherein said gray level with respect to said input region is equal to and/or smaller than said mean value.

9. The method of claim 6 further comprising:

defining a next most significant bit by repeating said procedure by taking said mean value of said input region pixel associated with said pixel set to 1 in said output image and similarly for said pixel set to 0; and

repeating said procedure to re-quantize said local image regions to any required bit depth.

10. The method of claim 1 further comprising automatically removing said local background variation to reveal an underlying pattern by said quantization transformation to efficiently manage a high dimensionality feature vector.

7. The method of dam 1 wherein said set of machine learning classifiers comprises a set of split up sparse network of winnows machine learning image classifiers.

11. A system for recognizing a license plate character, said system comprising:

a processor; and

a computer-usable medium embodying computer code, said computer-usable medium being coupled to said data bus, said computer program code comprising instructions executable by said processor and configured for;

capturing a license plate image with respect to a vehicle utilizing an image capturing unit and thereafter segmenting said license plate image into a license plate character image;

pre-processing said license plate character image to remove a local background variation with respect to said license plate character image and to define a local feature thereof utilizing a quantization transformation; and

identifying a classification margin for each character image utilizing a set of machine learning classifiers each binary in nature for said character image in order to thereafter determine a character type associated with at least one machine learning classifier among said set of machine learning classifiers with a largest classification margin to declare an optical character recognition result.

12. The system of claim 11 wherein said instructions are further configured for utilizing a multi-class machine learning classifier instead of said set of machine learning classifiers in order to declare said optical character recognition result.

13. The system of claim 11 wherein said instructions are further configured for training each binary classifier utilizing a character sample as a positive class and other characters as well as non-character images as a negative class.

14. The system of claim 11 wherein said instructions are further configured for:

grouping said license plate character image into an image training set of each character type; and

performing training by applying said quantization transformation and said set of machine learning classifiers for each training set and each character image.

15. The system of claim 11 wherein said quantization transformation comprises a successive mean quantization transformation.

16. The system of claim 11 wherein said set of machine learning classifiers comprises a set of sparse network of winnows machine based image classifiers.

17. The system of claim 11 wherein said set of machine learning classifiers comprises a set of split up sparse network of winnows machine learning image classifiers.

18. A non-transitory processor-readable medium storing code representing instructions to cause a processor to perform a process to recognize a license plate character, said code comprising code to:

capture a license plate image with respect to a vehicle utilizing an image-capturing unit and thereafter segmenting said license plate image into a license plate character image;

pre-process said license plate character image to remove a local background variation with respect to said license plate character image and to define a local feature thereof utilizing a quantization transformation; and

identify a classification margin for each character image utilizing a set of machine learning classifiers each binary in nature for said character image in order to thereafter determine a character type associated with at least one machine learning classifier among said set of machine learning classifiers with a largest classification margin to declare an optical character recognition result.

19. The non-transitory processor-readable medium of claim 18 wherein said code further comprises code to utilize a multi-class machine learning classifier instead of said set of machine learning classifiers in order to declare said optical character recognition result.

20. The non-transitory processor-readable medium of claim 18 wherein said set of machine learning classifiers comprises a set of split up sparse network of machine learning image classifiers.