Method and apparatus for identifying motor vehicles for monitoring traffic

• US 9,177,211 B2
• Filed: 08/23/2012
• Issued: 11/03/2015
• Est. Priority Date: 08/26/2011
• Status: Active Grant

• Current Assignee: Jenoptik Robot GmbH (Jenoptik AG)
• Original Assignee: Jenoptik Robot GmbH (Jenoptik AG)
• Inventors: Lehning, Michael
• Primary Examiner(s): KOSTAK, VICTOR R
• Application Number:
  US13/592,888
• Publication Number:
  US 20130050492A1
• Time in Patent Office: 1,167 Days
• Field of Search: 348/148, 348/149, 340/937, 340/933, 382/104, 382/105, 382/176, 382/182
• US Class Current: 1/1
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G06V 20/625 License plates

Abstract

A method and an apparatus for identifying motor vehicles for monitoring traffic. The identification is carried out by image-evaluation and includes determining the size ratios of a license-plate contour in a perspectively distorted image on the basis of stored standardized license-plate formats, determining the size of the perspective distortion of the license-plate contour on the basis of the associated standardized license-plate format, establishing a calculation rule for the perspective rectification on the basis of the ascertained distortion of the license-plate contour with respect to the associated license-plate format, rectifying the extracted license-plate-containing motor-vehicle view, and comparing the rectified image with reference images of front views of motor vehicles stored in a database in order to assign the image with greatest correspondence to a group of classified motor vehicles.

Claims

1. A method for identifying motor vehicles for monitoring traffic, comprising the following steps:

a) recording a perspective image of a motor vehicle under an acute angle with respect to the road, wherein the recorded image contains at least a motor-vehicle view that has a license plate,

b) recognizing the license plate in the recorded image and capturing a license-plate contour,

c) ascertaining the format of the license-plate contour from the comparison of the corner points of the license-plate contour with stored standardized license-plate formats from a memory,

d) ascertaining a calculation rule for perspective rectification of the license-plate contour on the basis of the associated standardized license-plate format,

e) producing an image region which is perspectively distorted in a matching manner with respect to the license-plate contour, with a prespecified magnification in dependence on the ascertained format of the license-plate contour,

f) extracting the content of the distorted image region surrounding the license-plate contour, and discarding the remaining recorded image of the motor vehicle,

g) adapting a license-plate-containing motor-vehicle view, which is extracted from the distorted surrounding image region, to the format of reference images stored in a database,

h) rectifying the motor-vehicle view using the calculation rule ascertained from the license-plate contour,

i) comparing similarities of the captured contours and their position with those of reference images from the database, and recognizing the greatest correspondence of the captured contours with one of the reference images, and

j) categorizing the recorded motor vehicle into a reference-image class, with which the reference image with the greatest correspondence was associated, and outputting a motor-vehicle class derived from the reference-image class.

2. The method as claimed in claim 1, wherein, in the production of the surrounding image region, its size ratio and position with respect to the license-plate contour are derived from the ascertained standardized license-plate format.

3. The method as claimed in claim 1, wherein, after extracting the surrounding distorted image region, characteristic features in the form of contours within the extracted license-plate-containing motor-vehicle view are captured.

4. The method as claimed in claim 1, wherein a correlation method is used for the similarity comparison of the captured contours with the stored reference images.

5. The method as claimed in claim 1, wherein the positions, distances and size ratios of the contours with respect to license-plate contour are compared for the similarity comparison of the captured contours with the stored reference images.

6. The method as claimed in claim 1, wherein the image recording takes place under a lateral angle with respect to the road and the motor vehicle is recorded in a perspective view.

7. The method as claimed in claim 1, wherein the image recording takes place under an elevation angle with respect to the road and the motor vehicle is recorded in a perspective view.

8. The method as claimed in claim 6, wherein the angle between the road and the image-capture device is selected to be at most so great that the license-plate-containing motor-vehicle view still has a sufficiently large area of the license-plate contour to allow a unique association of the standardized license-plate formats and thus the perspective rectification is independent from the knowledge of the angle of the image recording with respect to the road of the motor vehicle.

9. The method as claimed in claim 7, wherein the angle between the road and the image-capture device is selected to be at most so great that the license-plate-containing motor-vehicle view still has a sufficiently large area of the license-plate contour to allow a unique association of the standardized license-plate formats and thus the perspective rectification is independent from the knowledge of the angle of the image recording with respect to the road of the motor vehicle.

10. The method as claimed in claim 1, wherein, in addition to the license-plate contour, the position and size of the lettering of the license plate are captured for the ascertainment of the standardized license-plate format.

11. The method as claimed in claim 1, wherein, in addition to the license-plate contour, the registration number of the recorded license plate is captured, subjected to an OCR method and stored for the prosecution of a traffic violation.

12. The method as claimed in claim 11, wherein, in addition to the image recording, still further measurement data in connection with a traffic violation, date and time are simultaneously captured and stored with reference to the recorded image of the motor vehicle by saving this data in a script line directly in the image or in a data container within the image file.

13. An apparatus for identifying motor vehicles for monitoring traffic, comprising

an image-capture device being aligned for recording an image of the motor vehicle with at least one motor-vehicle view, which contains a license plate,

a database with reference images,

a computer based image-evaluation unit with memory containing,

computer software configured to recognize said license plate,

computer software configured to separate image background and to extract a relevant motor-vehicle view from the recorded image,

computer software configured to capture characteristic features within the extracted motor-vehicle view,

computer software configured to determine the size ratios of a license-plate contour in the perspectively distorted image on the basis of stored standardized license-plate formats,

computer software configured to determine the size of the perspective distortion of the license-plate contour on the basis of the associated standardized license-plate format,

computer software configured to establish a calculation rule for the perspective rectification on the basis of the ascertained distortion of the license-plate contour with respect to the associated license-plate format, and

computer software configured to rectify the extracted license-plate-containing motor-vehicle view, wherein the calculation rule for rectification of the license-plate contour is intended for the rectification of the entire extracted license-plate-containing motor-vehicle view such that the perspective rectification is achieved from the exact knowledge of the angle of the image recording with respect to the road of the motor vehicle;

an image comparison unit having

computer software configured to compare characteristic features of the extracted motor-vehicle view with those of reference images from a database,

computer software configured to recognize the greatest correspondence of the motor-vehicle view with one of the reference images and

computer software configured to associate the ascertained reference image with the greatest correspondence to a motor-vehicle class, and

an output unit.

14. The apparatus as claimed in claim 13,

wherein the image-evaluation unit has computer software configured to produce a surrounding image region, which is perspectively distorted in a matching manner with respect to the license-plate contour, for extracting the license-plate-containing motor-vehicle view to be evaluated from the recorded image, wherein said computer software configured to produce the surrounding image region contain an adjustment of the size of the image region with respect to the license-plate contour in dependence on the standardized license-plate format associated with the license-plate contour.

15. The apparatus as claimed in claim 14, wherein the image-evaluation unit has computer software configured to determine characteristic features in the form of contours within the produced surrounding image region, and the result from the computer software configured to determine the contours is supplied to the computer software configured to rectify the extracted license-plate-containing motor-vehicle view.

16. The apparatus as claimed in claim 13, wherein the apparatus additionally has a measurement system for determining the speed of the motor vehicle and computer software configured to recognize the registration number within the license-plate contour.