Geometric pre-correction for automatic license plate recognition

• US 9,020,200 B2
• Filed: 06/12/2012
• Issued: 04/28/2015
• Est. Priority Date: 06/12/2012
• Status: Active Grant

• Current Assignee: Conduent Business Services, LLC (Conduent, Inc.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Fan, Zhigang, Zhao, Yonghui, Wang, Shen-ge
• Primary Examiner(s): Koziol, Stephen R
• Assistant Examiner(s): SCHWARTZ, RAPHAEL M
• Application Number:
  US13/494,302
• Publication Number:
  US 20130329961A1
• Time in Patent Office: 1,050 Days
• Field of Search: None
• US Class Current: 382/105
• CPC Class Codes: G06T 5/80 Geometric correction
• CPC Class Codes: G06V 10/243 by compensating for image skew or non-uniform image deformations
• CPC Class Codes: G06V 20/63 Scene text, e.g. street names

Abstract

A method for adjusting a license plate that is detected in a captured image includes automatically determining at least one set of correction parameters corresponding to a slant-oriented license plate. The method further includes receiving an input image representing a detected license plate. In response to receiving the input image, the method includes automatically adjusting the input image to obtain a corrected image using the at least one set of correction parameters.

Claims

1. A method for adjusting a license plate detected in a captured image, the method comprising:

providing at least one set of correction parameters, the providing the at least one correction matrix including;

determining a feature set by calculating differences between the x,y coordinates of the four corners;

determining an original plate geometry matrix V;

of the feature set;

determining an ideal plate geometry D, wherein vectors of the D matrix are defined by a height and width of an ideal rectangular plate;

determining a corrected plate geometry CV.sub.i;

determining a distance between the corrected plate geometry and the ideal plate geometry by subtracting the ideal plate geometry from the corrected plate geometry (CV.sub.i−

D); and

,

determining the at least one set of correction parameters by squaring a sum of all matrix entries for the license plates;

receiving an input image representing a detected license plate;

in response to receiving the input image, automatically adjusting the input image to obtain a corrected image using the at least one correction matrix.

2. The method of claim 1, wherein the adjusting further includes:

determining if the slant-oriented plate was detected after applying the first set of correction parameters;

in response to the slant-oriented plate not being detected, applying at least a second set of correction parameters corresponding to an upright-oriented plate.

3. The method of claim 1, wherein the adjusting includes:

applying one set of correction parameters if the input image includes a front plate and applying two matrices if the input image includes a rear plate.

4. The method of claim 1, wherein the adjusting includes:

scaling the detected plate in the input image to a predetermined size; and

,

adjusting the detected plate in the input image to compensate for distortion.

5. The method of claim 1, wherein the providing the at least one set of correction parameters includes:

providing an image capture device;

receiving images of N number of license plates captured by the image capture device;

determining x,y coordinates for each of four corners of each license plate captured in the images;

determining a set of correction parameters for each image.

6. The method according to claim 1, wherein the providing the at least one set of correction parameters further includes:

determining a matrix V′

containing the plate information for the N number of plates;

determining an ideal plate geometry D′

; and

,

calculating vector c′

using the matrix V′ and

matrix D′

.

7. The method according to claim 6, wherein the N number of plates all include one of upright-oriented and slant-oriented plates.

8. The method according to claim 6, wherein the determining the ideal plate geometry D′

includes;

specifying an ideal height and width of the number N of license plates.

9. A computer program product comprising non-transitory media which encodes instructions for performing the method of claim 1.

10. A system for adjusting an input image, the system comprising:

a correction device comprising memory which stores instructions for performing the method of claim 1 and a processor, in communication with the memory for executing the instructions.

11. A system for adjusting an input image, the system comprising:

an input device adapted to capture an image of a license plate;

a correction device in communication with the input device, the correction device adapted to;

receive the input image transmitted from the input device

determine x,y coordinates for each of four corners of the license plate in the input image;

determine a feature set by calculating differences between the x,y coordinates of the four corners;

determine an original plate geometry matrix V_i of the feature set;

determine an ideal plate geometry D, wherein vectors of the D matrix are defined by a height and width of an ideal rectangular plate;

determine a corrected plate geometry CV_i;

determine a distance between the corrected plate geometry and the ideal plate geometry by subtracting the ideal plate geometry from the corrected plate geometry (CV_i−

D);

obtain an optimization using a least square root method; and

in response to receiving the input image, automatically adjusting the input image v[(x′

,y′

)] to obtain a corrected image v′

(x,y) by applying at least one correction matrix;

a storage device in communication with the correction device, the storage device adapted to store the at least one correction matrix;

an interface in communication with the correction device, the interface adapted to display the corrected image.

12. The system according to claim 11, further comprising:

a training device in communication with the storage device, the training device adapted to;

determine a matrix V′

containing the plate information for the N number of plates;

determine an ideal plate geometry D′

; and

,

calculate vector c′

using the matrix V′ and

matrix D′

.

13. The system according to claim 12, wherein the N number of plates includes one of front plates, rear slant-oriented plates, and rear upright-oriented plates.

14. The system according to claim 11, wherein the at least one correction matrix scales the image and corrects for distortion caused by the input device.

15. The system according to claim 11, wherein the correction device determines the four corners by detecting boundaries of the license plate and calculating intersections of the boundaries.

16. The system according to claim 11, wherein the at least one correction matrix corresponds to a slant-oriented license plate.

17. The system according to claim 16, wherein the storage device is adapted to store at least a second correction matrix corresponding to an upright-oriented license plate and the correction device is adapted to automatically adjust the input image by applying the at least second correction matrix after applying the at least first correction matrix.

18. A system for adjusting an input image, the system comprising:

a training device adapted to;

determine a feature set by calculating differences between the x,y coordinates of the four corners;

determine an original plate geometry matrix V_i of the feature set;

determine an ideal plate geometry D, wherein vectors of the D matrix are defined by a height and width of an ideal rectangular plate;

determine a corrected plate geometry CV_i;

determine a distance between the corrected plate geometry and the ideal plate geometry by subtracting the ideal plate geometry from the corrected plate geometry (CV_i−

D); and

,

determine a correction matrix C by squaring a sum of all matrix entries for the license plates;

a storage device in communication with the training device, the storage device adapted to store the at least one correction matrix;

an input device adapted to capture an image of a detected license plate;

a correction device in communication with the input device and the storage device, the correction device adapted to;

receive the captured image transmitted from the input device, and in response to receiving the input image, automatically adjusting the input image v[(x′

,y′

)] to obtain a corrected image v′

(x,y) by applying the at least one correction matrix;

an interface in communication with the correction device, the interface adapted to display the corrected image.