METHODS AND SYSTEMS FOR IMPROVED LICENSE PLATE SIGNATURE MATCHING BY SIMILARITY LEARNING ON SYNTHETIC IMAGES

• US 20130129151A1
• Filed: 11/18/2011
• Published: 05/23/2013
• Est. Priority Date: 11/18/2011
• Status: Active Grant

• Current Assignee: Conduent Business Services, LLC (Conduent, Inc.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: RODRIGUEZ SERRANO, Jose Antonio, BALA, Raja, PERRONNIN, Florent, ZHAO, Yonghui, SAUNDERS, Craig John
• Granted Patent: US 8,588,470 B2
• Time in Patent Office: Days
• US Class Current: 382/105
• CPC Class Codes: G06F 18/21 Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 10/70 using pattern recognition or machine learning optical pattern recognition or electronic computations therefor G06V10/88
• CPC Class Codes: G06V 10/772 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries
• CPC Class Codes: G06V 20/62 Text, e.g. of license plates, overlay texts or captions on TV images
• CPC Class Codes: G06V 20/625 License plates

Abstract

Methods and systems for improved license plate signature matching by similarity learning on synthetic images comprise generating a plurality of synthetic license plate images; applying one or more transformations to the synthetic license plate images to cause the synthetic license plate images to more closely resemble authentic license plate image captures; and providing the synthetic license plate images as inputs to a machine distance learning algorithm in which weighted similarity scores are calculated between signatures of analogous and non-analogous license plate images and one or more sets of signature weights are iteratively adjusted to increase the likelihood that comparing analogous license plate images results in high weighted signature similarity scores and comparing non-analogous license plate images results in low weighted signature similarity scores.

Claims

1. A method of implementing license plate signature matching by similarity learning on synthetic images, the method comprising:

generating a plurality of synthetic license plate images;

selecting a plurality of image pairs comprising a first synthetic license plate image associated with known first identification information and a second license plate image associated with known second identification information;

for each image pair in the plurality of image pairs;

generating a first signature representing the first synthetic license plate image;

generating a second signature representing the second license plate image;

calculating a weighted similarity score reflecting a degree of similarity between the first signature and the second signature using a set of one or more signature weights; and

adjusting the set of one or more signature weights to increase a likelihood that the weighted similarity score reflects a degree of similarity between the first identification information and the second identification information.

2. The method of claim 1, wherein the second license plate image is a synthetic license plate image.

3. The method of claim 1, wherein generating the plurality of synthetic license plate images comprises:

retrieving a blank license plate template;

retrieving a plurality of sequences of license plate characters to be rendered onto the template;

obtaining character font, size, and color characteristics associated with the template; and

rendering each sequence of characters onto the template in conformance with the obtained character font, size, and color characteristics.

4. The method of claim 3 wherein one or more of the character font, size, and color characteristics are derived by analysis of authentic license plate images having the template.

5. The method of claim 1, wherein generating the plurality of synthetic license plate images comprises:

applying one or more transformations to one or more synthetic license plate images to cause the one or more synthetic license plate images to resemble one or more authentic license plate image captures.

6. The method of claim 1, wherein one or more of the first signature and the second signature is a Fisher vector.

7. The method of claim 1, wherein one or more of the first identification information and the second identification information comprises a license plate number.

8. The method of claim 7, wherein the one or more of the first identification information and the second identification information further comprises an issuing authority.

9. The method of claim 1, further comprising:

receiving an authentic license plate image;

calculating a signature of the authentic license plate image;

locating a matching synthetic license plate image from among the plurality of synthetic license plate images by comparing the signature of the authentic license plate image to signatures of one or more of the plurality of synthetic license plate images using the adjusted set of one or more signature weights; and

determining identification information associated with the authentic license plate image by extracting known identification information associated with the matching synthetic license plate image.

10. A method of identifying metadata associated with a target vehicle, the method comprising:

receiving identification information associated with a target vehicle;

generating a synthetic license plate image using the identification information;

calculating a signature of the synthetic license plate image;

locating a matching authentic license plate image by matching the signature of the authentic license plate image to the signature of the synthetic license plate image; and

extracting metadata associated with the authentic license plate image.

11. The method of claim 10, wherein generating the synthetic license plate image comprises:

applying one or more transformations to the synthetic license plate image to cause the synthetic license plate image to resemble an authentic license plate image.

12. The method of claim 10, wherein the metadata comprises a time when the authentic license plate image was captured.

13. The method of claim 10, wherein the metadata comprises a place where the authentic license plate image was captured.

14. A system configured to implement license plate signature matching by similarity learning on synthetic images, the system comprising:

a processing system comprising one or more processors; and

a memory system comprising one or more computer-readable media, wherein the computer-readable media contain instructions stored thereon that, when executed by the processing system, cause the processing system to perform operations comprising;

generating a plurality of synthetic license plate images;

selecting a plurality of image pairs comprising a first synthetic license plate image associated with known first identification information and a second license plate image associated with known second identification information;

for each image pair in the plurality of image pairs;

generating a first signature representing the first synthetic license plate image;

generating a second signature representing the second license plate image;

calculating a weighted similarity score reflecting a degree of similarity between the first signature and the second signature using a set of one or more signature weights; and

adjusting the set of one or more signature weights to increase a likelihood that the weighted similarity score reflects a degree between the first identification information and the second identification information.

15. The system of claim 14, wherein the second license plate image is a synthetic license plate image.

16. The system of claim 14, wherein generating the plurality of synthetic license plate images comprises:

retrieving a blank license plate template;

retrieving a plurality of sequences of license plate characters to be rendered onto the template;

obtaining character font, size, and color characteristics associated with the template; and

rendering each sequence of characters onto the template in conformance with the obtained character font, size, and color characteristics.

17. The system of claim 16, wherein one or more of the character font, size, and color characteristics are derived by analysis of authentic license plate images having the template.

18. The system of claim 14, wherein generating the plurality of synthetic license plate images comprises:

applying one or more transformations to one or more synthetic license plate images to cause the one or more synthetic license plate images to resemble one or more authentic, license plate image captures.

19. The system of claim 14, wherein one or more of the first signature and the second signature is a Fisher vector.

20. The system of claim 14, wherein one or more of the first identification information and the second identification information comprises a license plate number.

21. The system of claim 20, wherein the one or more of the first identification information and the second identification information further comprises an issuing authority.

22. The system of claim 14, the operations further comprising:

receiving an authentic license plate image;

calculating a signature of the authentic license plate image;

locating a matching license plate image from among the plurality of license plate images by comparing the signature of the authentic license plate image to signatures of one or more of the plurality of synthetic license plate images using the adjusted set of one or more signature weights; and

determining identification information associated with the authentic license plate image by extracting known identification information associated with the matching synthetic license plate image.

23. A system configured to identify metadata associated with a target vehicle, the system comprising:

a processing system comprising one or more processors; and

a memory system comprising one or more computer-readable media, wherein the computer-readable media contain instructions stored thereon that, when executed by the processing system, cause the processing system to perform operations comprising;

receiving identification information associated with a target vehicle;

generating a synthetic license plate image using the identification information;

calculating a signature of the synthetic license plate image;

locating a matching authentic license plate image by matching the signature of the authentic license plate image to the signature of the synthetic license plate image; and

extracting metadata associated with the authentic license plate image.

24. The system of claim 23, wherein generating the synthetic license plate image comprises:

applying one or more transformations to the synthetic license plate image to cause the synthetic license plate image to resemble an authentic license plate image.

25. The system of claim 23, wherein the metadata comprises one of a time when and a place where the authentic license plate image was captured.