Vehicle vision system with object detection

• US 10,423,842 B2
• Filed: 02/04/2019
• Issued: 09/24/2019
• Est. Priority Date: 07/30/2015
• Status: Active Grant

• Current Assignee: Magna Electronics Incorporated (Magna International Incorporated)
• Original Assignee: Magna Electronics Incorporated (Magna International Incorporated)
• Inventors: Gupta, Nikhil, Zhang, Liang
• Primary Examiner(s): Saini, Amandeep
• Application Number:
  US16/266,180
• Publication Number:
  US 20190171889A1
• Time in Patent Office: 232 Days
• Field of Search: 382104
• CPC Class Codes: G06T 2207/30261 Obstacle
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: G06V 20/58 Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads

Abstract

A vision system of a vehicle includes at least one camera disposed at a vehicle and having a field of view exterior of the vehicle, and an image processor operable to process image data captured by the camera. Responsive to image processing of captured image data, the image processor determines objects present in the field of view of the camera. The vision system processes additional frames of captured image data to enhance determination of objects of interest. The vision system initially detects an object present in the field of view of the camera and conducts hypotheses filtering and hypotheses merging and, responsive to the hypotheses merging, the system determines that the detected object is an object of interest or determines that the detected object is not an object of interest.

Claims

1. A vision system for a vehicle, said vision system comprising:

at least one camera disposed at a vehicle and having a field of view exterior of the vehicle;

wherein said camera comprises a pixelated imaging array having a plurality of photosensing elements;

a control having an image processor that processes image data captured by said camera;

wherein, responsive to processing by said image processor of captured image data, objects present in the field of view of said camera are detected;

wherein, after detection via processing of captured image data by said image processor of at least one object present in the field of view of said camera, additional frames of captured image data are processed to discern the detected object being an object of interest from the detected object being not an object of interest;

wherein said vision system utilizes hypotheses filtering and hypotheses merging to discern that the detected object is an object of interest;

wherein hypotheses filtering comprises detecting the detected object in an initial frame of captured image data and tracking the detected object over multiple sets of frames of image data captured subsequent to the initial frame of captured image data;

wherein hypotheses merging comprises comparing outputs of hypotheses filtering for multiple sets of frames of captured image data; and

wherein, responsive to the results of hypotheses merging, said vision system determines that the detected object is an object of interest or determines that the detected object is not an object of interest.

2. The vision system of claim 1, wherein said vision system predicts hypotheses and evaluates predicted hypotheses to determine that the detected object is an object of interest or to determine that the detected object is not an object of interest.

3. The vision system of claim 2, wherein determination that the detected object is an object of interest by said vision system comprises hypotheses evaluation based at least in part on the predicted hypotheses.

4. The vision system of claim 1, wherein determination that the detected object is an object of interest by said vision system comprises hypotheses prediction based at least in part on the results of hypotheses merging.

5. The vision system of claim 4, wherein said vision system repeats the hypotheses merging and the hypotheses prediction to enhance determination that the detected object is an object of interest.

6. The vision system of claim 4, wherein filtered hypotheses and predicted hypotheses are combined to generate assigned hypotheses.

7. The vision system of claim 6, wherein the assigned hypotheses are input into a hypotheses prediction module, and wherein an output of the hypotheses prediction module is (i) input to hypotheses merging and (ii) input into a hypotheses evaluation module.

8. The vision system of claim 1, wherein, responsive to the results of hypotheses merging, said vision system determines that the detected object is a pedestrian and not an object not of interest.

9. The vision system of claim 8, wherein, responsive to the results of hypotheses merging, said vision system discerns the pedestrian from a tree.

10. The vision system of claim 1, wherein, responsive to the results of hypotheses merging, said vision system determines that the detected object is a bicycle and not an object not of interest.

11. The vision system of claim 10, wherein, responsive to the results of hypotheses merging, said vision system discerns the bicycle from a tree.

12. The vision system of claim 1, wherein hypotheses merging and hypotheses prediction comprises use of Kalman filtering.

13. The vision system of claim 1, wherein said camera comprises a fisheye lens.

14. A vision system for a vehicle, said vision system comprising:

at least one camera disposed at a vehicle and having a field of view exterior of the vehicle;

wherein said camera comprises a pixelated imaging array having a plurality of photosensing elements;

a control having an image processor that processes image data captured by said camera;

wherein, responsive to processing by said image processor of captured image data, objects present in the field of view of said camera are detected;

wherein, after detection via processing of captured image data by said image processor of at least one object present in the field of view of said camera, additional frames of captured image data are processed to discern the detected object being a pedestrian from the detected object being not a pedestrian;

wherein said vision system utilizes hypotheses filtering and hypotheses merging to discern that the detected object is a pedestrian;

wherein hypotheses filtering comprises detecting the detected object in an initial frame of captured image data and tracking the detected object over multiple sets of frames of image data captured subsequent to the initial frame of captured image data;

wherein hypotheses merging comprises comparing outputs of hypotheses filtering for multiple sets of frames of captured image data;

wherein hypotheses merging and hypotheses prediction comprises use of Kalman filtering; and

wherein, responsive to the results of hypotheses merging, said vision system determines that the detected object is a pedestrian or determines that the detected object is not a pedestrian.

15. The vision system of claim 14, wherein, responsive to the results of hypotheses merging, said vision system discerns the pedestrian from a tree.

16. The vision system of claim 14, wherein determination that the detected object is a pedestrian by said vision system comprises hypotheses prediction based at least in part on the results of hypotheses merging, and wherein said vision system repeats the hypotheses merging and the hypotheses prediction to enhance determination that the detected object is a pedestrian.

17. The vision system of claim 16, wherein filtered hypotheses and predicted hypotheses are combined to generate assigned hypotheses, and wherein the assigned hypotheses are input into a hypotheses prediction module, and wherein an output of the hypotheses prediction module is (i) input to hypotheses merging and (ii) input into a hypotheses evaluation module.

18. A vision system for a vehicle, said vision system comprising:

at least one camera disposed at a vehicle and having a field of view exterior of the vehicle;

wherein said camera comprises a pixelated imaging array having a plurality of photosensing elements;

a control having an image processor that processes image data captured by said camera;

wherein, responsive to processing by said image processor of captured image data, objects present in the field of view of said camera are detected;

wherein, after detection via processing of captured image data by said image processor of at least one object present in the field of view of said camera, additional frames of captured image data are processed to discern the detected object being a bicycle from the detected object being not a bicycle;

wherein said vision system utilizes hypotheses filtering and hypotheses merging to discern that the detected object is a bicycle;

wherein hypotheses filtering comprises detecting the detected object in an initial frame of captured image data and tracking the detected object over multiple sets of frames of image data captured subsequent to the initial frame of captured image data;

wherein hypotheses merging comprises comparing outputs of hypotheses filtering for multiple sets of frames of captured image data;

wherein hypotheses merging and hypotheses prediction comprises use of Kalman filtering; and

wherein, responsive to the results of hypotheses merging, said vision system determines that the detected object is a bicycle or determines that the detected object is not a bicycle.

19. The vision system of claim 18, wherein, responsive to the results of hypotheses merging, said vision system discerns the bicycle from a tree.

20. The vision system of claim 18, wherein determination that the detected object is a bicycle by said vision system comprises hypotheses prediction based at least in part on the results of hypotheses merging, and wherein said vision system repeats the hypotheses merging and the hypotheses prediction to enhance determination that the detected object is a bicycle.

21. The vision system of claim 20, wherein filtered hypotheses and predicted hypotheses are combined to generate assigned hypotheses, and wherein the assigned hypotheses are input into a hypotheses prediction module, and wherein an output of the hypotheses prediction module is (i) input to hypotheses merging and (ii) input into a hypotheses evaluation module.