System and method for detecting features from images of vehicles

• US 20060177100A1
• Filed: 12/21/2005
• Published: 08/10/2006
• Est. Priority Date: 02/09/2005
• Status: Active Grant

• Current Assignee: Continental Automotive GmbH (Continental AG), Siemens Corporation USA (Siemens AG)
• Original Assignee: Siemens Corp. (Siemens AG)
• Inventors: Ramesh, Visvanathan, Zhu, Ying, Kohler, Thorsten, Pellkofer, Martin, Xie, Binglong
• Granted Patent: US 7,620,208 B2
• Time in Patent Office: Days
• US Class Current: 382/104
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 2201/08 Detecting or categorising vehicles

Abstract

A method of detecting a feature of a vehicle in an image of a vehicle includes providing a digitized image of a vehicle, providing a first filter mask over a first subdomain of the image, where the filter mask is placed to detect a feature in the image, calculating a function of a gradient of the image inside the first masked subdomain, and detecting the presence or absence of a vehicle feature within the first masked subdomain based on the value of the gradient function.

Claims

1. A method of detecting a feature of a vehicle in an image of said vehicle, said method comprising the steps of:

providing a digitized image of said vehicle, said image comprising a plurality of intensities corresponding to a domain of points on an 2-dimensional grid;

providing a first filter mask over a first subdomain of said image, wherein said filter mask is placed to detect a feature in said image;

calculating a function of a gradient of said image inside said first masked subdomain; and

detecting the presence or absence of a vehicle feature within said first masked subdomain based on the value of said gradient function.

2. The method of claim 1, wherein said vehicle is an automobile.

3. The method of claim 1, wherein said vehicle is a truck.

4. The method of claim 1, wherein said function is a sum of an absolute value of the log gradient over all pixels in said first masked subdomain.

5. The method of claim 1, further comprising varying the size and position of said first filter mask to obtain a set of said features.

6. The method of claim 4, further comprising providing a second filter mask over a second subdomain of said image, wherein said function is a sum of an absolute value of the log gradient over all pixels in said second masked subdomain.

7. The method of claim 6, wherein said feature is detected by analyzing a difference between a sum of a component of said gradient function over said first subdomain and a sum of a same component of said gradient function over said second subdomain.

8. The method of claim 6, wherein said feature is detected by analyzing a ratio of a sum of a component of said gradient function over said first subdomain and a sum of a same component of said gradient function over said second subdomain.

9. The method of claim 6, wherein said feature is detected by analyzing a comparison of a sum of a component of said gradient function over said first subdomain with a sum of a same component of said gradient function over said second subdomain.

10. The method of claim 4, wherein said feature is detected by analyzing a sum of different components of said gradient function over said first subdomain.

11. The method of claim 4, wherein said feature is detected by analyzing a ratio of different components of said gradient function over said first subdomain.

12. The method of claim 6, wherein the size and position of said second filter mask are varied to obtain a set of said features.

13. A method of detecting a feature of a vehicle in an image of said vehicle, said method comprising the steps of:

providing a digitized image of said vehicle, said image comprising a plurality of intensities corresponding to a domain of points on an 2-dimensional grid;

providing at least one filter mask over a subdomain of said image, wherein said filter mask is placed to detect a feature in said image, wherein the size and position of said at least one filter mask are varied to obtain a set of said features; and

calculating a sum of an absolute value of the log gradient over all pixels in said masked subdomain.

14. The method of claim 13, further comprising detecting the presence or absence of a vehicle feature within said masked subdomain based on the value of said sum of said absolute value of said log gradient.

15. A program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform the method steps for detecting a feature of a vehicle in an image of said vehicle, said method comprising the steps of:

providing a digitized image of said vehicle, said image comprising a plurality of intensities corresponding to a domain of points on an 2-dimensional grid;

providing a first filter mask over a first subdomain of said image, wherein said filter mask is placed to detect a feature in said image;

calculating a function of a gradient of said image inside said first masked subdomain; and

detecting the presence or absence of a vehicle feature within said first masked subdomain based on the value of said gradient function.

16. The computer readable program storage device of claim 15, wherein said vehicle is an automobile.

17. The computer readable program storage device of claim 15, wherein said vehicle is a truck.

18. The computer readable program storage device of claim 15, wherein said function is a sum of an absolute value of the log gradient over all pixels in said first masked subdomain.

19. The computer readable program storage device of claim 15, the method further comprising varying the size and position of said first filter mask to obtain a set of said features.

20. The computer readable program storage device of claim 18, the method further comprising providing a second filter mask over a second subdomain of said image, wherein said function is a sum of an absolute value of the log gradient over all pixels in said second masked subdomain.

21. The computer readable program storage device of claim 20, wherein said feature is detected by analyzing a difference between a sum of a component of said gradient function over said first subdomain and a sum of a same component of said gradient function over said second subdomain.

22. The computer readable program storage device of claim 20, wherein said feature is detected by analyzing a ratio of a sum of a component of said gradient function over said first subdomain and a sum of a same component of said gradient function over said second subdomain.

23. The computer readable program storage device of claim 20, wherein said feature is detected by analyzing a comparison of a sum of a component of said gradient function over said first subdomain with a sum of a same component of said gradient function over said second subdomain.

24. The computer readable program storage device of claim 18, wherein said feature is detected by analyzing a sum of different components of said gradient function over said first subdomain.

25. The computer readable program storage device of claim 18, wherein said feature is detected by analyzing a ratio of different components of said gradient function over said first subdomain.

26. The computer readable program storage device of claim 20, wherein the size and position of said second filter mask are varied to obtain a set of said features.