Object detection and recognition system

• US 8,131,018 B2
• Filed: 02/08/2008
• Issued: 03/06/2012
• Est. Priority Date: 02/08/2008
• Status: Active Grant

• Current Assignee: Joyson Safety Systems Acquisition, LLC
• Original Assignee: TK Holdings Incorporated (Ningbo Junsheng Investment Group Co., Ltd.)
• Inventors: Wilson, Kitchener Clark
• Primary Examiner(s): Tabatabai, Abolfazl
• Application Number:
  US12/068,651
• Publication Number:
  US 20090202107A1
• Time in Patent Office: 1,488 Days
• Field of Search: 382/103, 382/104, 382/107, 382/236, 340/435, 340/557, 340/905, 340/988, 340/990, 340/901, 340/902, 340/933, 701/28, 701/36, 701/45, 701/93, 701/200, 701/210, 701/201, 348/154, 348/155, 348/169, 348/208.16, 348/352
• US Class Current: 382/104
• CPC Class Codes: G06V 20/582 of traffic signs

Abstract

An object recognition system is provided including at least one image capturing device configured to capture at least one image, wherein the image includes a plurality of pixels and is represented in an image data set, an object detection device configured to identify a plurality of pixels corresponding to objects from the at least one image, wherein an object includes a plurality of pixels and is represented in an object data set, wherein the object data set includes a set of features corresponding to each pixel in the object, and an image recognition device configured to recognize objects of interest present in an object by image correlation against a set of template images to recognize an object as one of the templates.

Claims

1. An object detection and recognition system for a vehicle, comprising:

at least one image capture device;

a vision controller configured to receive images from the image capture device;

an object detector configured to recognize an object of interest within the received images; and

a vehicle controller for receiving the object of interest and configured to control a vehicle subsystem in response to recognition of the object of interest,

wherein the object detector comprises an object segmenter configured to isolate individual objects within the received images and an object recognizer configured to determine whether the individual objects include the object of interest by curve fitting at least one object of interest template to one of the isolated individual objects.

2. An object detection and recognition system for a vehicle, as claimed in claim 1, wherein the vehicle subsystem is a braking system.

3. An object detection and recognition system for a vehicle, as claimed in claim 1, wherein the vehicle subsystem is a driver display.

4. An object detection and recognition system for a vehicle, as claimed in claim 1, wherein the object detector further comprises a database configured to store the at least one object of interest template.

5. An object detection and recognition system, as claimed in claim 4, wherein the object recognizer is configured to receive the at least one object of interest template from the database, and is configured to curve fit the at least one object of interest template to the one of the isolated individual objects by mapping the at least one object of interest template onto the one of the isolated individual objects by determining values of mapping parameters that minimize a function of deviations between the mapped template and the one of the isolated individual objects.

6. An object detection and recognition system, as claimed in claim 5, wherein the mapping parameters include a geometrical scale factor and an offset.

7. An object detection and recognition system, as claimed in claim 6, wherein the mapping parameters further include an illumination scale factor for each color component forming the one of the isolated individual objects.

8. An object detection and recognition system, as claimed in claim 5, wherein the at least one object of interest template received by the object recognizer is a plurality of object of interest templates from the database,

wherein the object recognizer is configured to:

pre-screen the object of interest templates to obtain pre-screen templates;

curve fit the pre-screened templates to the individual objects to obtain curve fitted templates;

calculate correlation between the curve fitted templates and the individual objects;

select an object of interest template with a maximum correlation; and

report the selected object of interest template to the vehicle controller if the maximum correlation exceeds a minimum value.

9. An object detection and recognition system, as claimed in claim 5, wherein GPS position data is used to enable or disable the at least one object of interest template received from the database.

10. An object detection and recognition system, as claimed in claim 1, wherein the object segmenter is configured to receive images from the at least one image capture device and isolate the individual objects within the received images.

11. An object detection and recognition system, as claimed in claim 10, wherein the object segmenter is configured to use color information to isolate the individual objects within the received images.

12. An object detection and recognition system, as claimed in claim 1, wherein the object segmenter is configured to receive images from at least two image capture devices to form a stereoscopic representation of the received images to obtain range information for the received images and configured to isolate the individual objects within the received images using said range information.

13. An object detection and recognition system, as claimed in claim 12, wherein the object segmenter further uses color information in combination with the range information to isolate the individual objects within the received images.

14. A method for object detection and recognition in a vehicle, comprising the steps of:

receiving captured images from at least one image capture device;

isolating individual objects within the received images;

determining whether the individual objects include an object of interest by curve fitting object of interest templates to the individual objects; and

controlling a vehicle subsystem in response to the object of interest.

15. A method for object detection and recognition in a vehicle, as claimed in claim 14, wherein the vehicle subsystem is a braking system.

16. A method for object detection and recognition in a vehicle, as claimed in claim 14, wherein the vehicle subsystem is a driver display.

17. A method for object detection and recognition in a vehicle, as claimed in claim 14, wherein the object of interest templates are stored in a database.

18. A method for object detection and recognition, as claimed in claim 17, wherein the curve fitting comprises mapping the object of interest templates onto the individual objects by determining values of mapping parameters that minimize a function of the deviations between the mapped templates and the individual objects.

19. A method for object detection and recognition, as claimed in claim 18, wherein the mapping parameters include a geometrical scale factor and an offset.

20. A method for object detection and recognition system, as claimed in claim 19, wherein the mapping parameters for each mapped template further include an illumination scale factor for, at each color component, forming its respective isolated individual object.

21. A method for object detection and recognition in a vehicle, as claimed in claim 17, wherein the determining step further comprises:

receiving the object of interest templates from the database;

pre-screening the object of interest templates to obtain pre-screen templates;

curve fitting the pre-screened templates to the individual objects to obtain curve fitted templates;

calculating correlation between the curve fitted templates and the individual objects;

selecting an object of interest template with a maximum correlation; and

reporting the selected object of interest template to the vehicle controller if the maximum correlation exceeds a minimum value.

22. A method for object detection and recognition in a vehicle, as claimed in claim 21, wherein GPS position data is used to enable or disable the object of interest templates received from the database.

23. A method for object detection and recognition in a vehicle, as claimed in claim 14, wherein the isolating step uses color information to isolate the individual objects within the received captured images.

24. A method for object detection and recognition in a vehicle, as claimed in claim 14, wherein the isolating step receives images from at least two image capture devices to form a stereoscopic representation of the received captured images to obtain range information for the received captured images and isolates the individual objects within the received captured images using said range information.

25. A method for object detection and recognition in a vehicle, as claimed in claim 24, wherein the isolating step uses color information in combination with the range information to isolate the individual objects within the received images.