Rules-based occupant classification system for airbag deployment

• US 6,459,974 B1
• Filed: 05/30/2001
• Issued: 10/01/2002
• Est. Priority Date: 05/30/2001
• Status: Expired due to Fees

• Current Assignee: Eaton Corp. (Wisconsin) (Eaton Corporation PLC.)
• Original Assignee: Eaton Corp. (Wisconsin) (Eaton Corporation PLC.)
• Inventors: Asif, Muhammad, Baloch, Aijaz Ahmed, Afridi, Khurram Khan, Ahmed, Usman
• Primary Examiner(s): ARTHUR JEANGLAUDE, GERTRUDE
• Application Number:
  US09/870,151
• Time in Patent Office: 489 Days
• Field of Search: 701/45, 701/47, 180/271, 180/272, 280/734, 280/735
• US Class Current: 701/45
• CPC Class Codes: B60R 21/01538 for image processing, e.g. cameras or sensor arrays
• CPC Class Codes: B60R 21/01556 Child-seat detection systems
• CPC Class Codes: G06F 18/24765 Rule-based classification
• CPC Class Codes: G06T 7/215 Motion-based segmentation
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: G06V 10/765 using rules for classification or partitioning the feature space
• CPC Class Codes: G06V 20/593 Recognising seat occupancy
• CPC Class Codes: G06V 40/10 Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands

Abstract

An occupant classification system utilizes a rules-based expert system to automatically classify the occupant of a seat for the purposes of airbag deployment. The invention provides users with the ability to create, test, and modify the image attributes or “features” used by the expert system to classify occupants into one of several predefined occupant-type categories. Users are also provided the ability to create, test, and modify the processes utilizing those chosen features. The user of the invention designs the features and the algorithms used by the expert system classifier. A feature extractor is used to extract features from an image of the occupant and surrounding seat area, and the values relating to those features are sent in a vector of features to the expert system classifier. The expert system classifier classifies the image of the occupant according to the internal rules for that classifier. The resulting occupant-type classification is sent to the confidence factor extractor, along with the vector of features. The confidence factor extractor generates a confidence factor indicating the probable accuracy of the occupant-type classification. The occupant-type classification and confidence factor are then sent to the airbag controller so the airbag deployment system can take the appropriate action. For embodiments involving multiple expert system classifiers, one weighted occupant-type classification and one weighted confidence factor are sent to the airbag controller.

Claims

1. An occupant classification system for use with an airbag deployment system having a seat, an occupant in the seat, a video camera for capturing a visual image of the occupant, a feature extractor, and an airbag controller, said occupant classification system comprising:

an expert system classifier, wherein said expert system classifier receives a vector of features comprising at least one feature relating to the occupant, and wherein said expert system classifier applies an internal set of rules to the vector of features such that the occupant is identified as one of a plurality of predefined occupant-type categories; and

a confidence factor extractor, wherein said confidence factor extractor generates a confidence factor representing the relative accuracy or inaccuracy of the occupant-type classification by said expert system classifier.

2. The occupant classification system in claim 1, wherein said predetermined occupant-type categories comprise adults, children, rear facing infant seat, front facing child seat, and booster seat.

3. The occupant classification system in claim 1, wherein the predictive qualities of each said feature is tested before it is used by said occupant classification system.

4. The occupant classification system in claim 1, wherein said vector of features is comprised of a plurality of features.

5. The occupant classification system in claim 1, wherein said occupant classification system is comprised of a plurality of said expert system classifiers.

6. The occupant classification system in claim 5, wherein all said expert system classifiers utilize the same set of said features.

7. The occupant classification system in claim 5, wherein each said expert system classifier utilizes said features unique to that said classifier.

8. The occupant classification system in claim 5, wherein a weighted voting scheme is used to generate a single occupant classification determination and a single confidence factor relating to that determination.

9. The occupant classification system in claim 1, wherein new features can be derived from old features.

10. The occupant classification system in claim 1, wherein a reference classifier is used to test and evaluate the predictive qualities of each feature used in said reference classifier before any said feature is used in said expert system classifier.

11. The occupant classification system in claim 10, wherein only those features rated average or higher at predicting occupant-types classifications have been included said set of features used by said expert system classifier.

12. The occupant classification system in claim 1, wherein data-mining techniques are used to maximize the predictive nature of said features.

13. The occupant classification system of claim 1, wherein there is only one video camera used with said occupant classification system.

14. The occupant classification system of claim 1, wherein said vector of features includes only features derived from a segmented image of the occupant.

15. The occupant classification system of claim 1, wherein said vector of features does not include a location feature.

16. The occupant classification system of claim 15, wherein said vector of features does not include a motion feature.

17. An occupant classification system comprising:

a sensor for capturing an image of a seat area, said visual image being in a frequency range normally perceivable by the human eye;

a feature extractor for extracting a feature of an occupant from said image and generating a vector of said feature, wherein said vector of features are derived solely from said visual image;

an expert system classifier that receives said vector and determines a classification of the occupant; and

a confidence factor extractor that uses said vector and said occupant-type classification to calculate a confidence level of said classification.

18. The occupant classification system in claim 17, wherein segmented image and ambient image are used to extract said features for said vector of features.

19. The occupant classification system in claim 17, wherein said expert system classifier classifies occupants into one of a plurality of predefined occupant-type classifications.

20. The occupant classification system in claim 19, wherein said classification types comprise the categories of adult, child, rear facing infant seat, front facing child seat, booster seat, miscellaneous, and indeterminate.

21. The occupant classification system in claim 17, wherein one or more features can be used to derive additional features.

22. The occupant classification system in claim 21, wherein new features are derived by adding two said features together, subtracting one said feature from another said feature value, multiplying two said features together, or the dividing one said feature by another said feature.

23. The occupant classification system in claim 17, wherein trigonometric functions can be used to derive additional features.

24. The occupant classification system in claim 17, wherein the mean value of a feature for a particular occupant-type classification can be used to derive additional features.

25. The occupant classification system in claim 17, wherein said vector of features comprises more than one feature.

26. The occupant classification system in claim 17, wherein said occupant classification system comprises more than one said expert classifier.

27. The occupant classification system in claim 26 wherein each said expert system classifier utilizes the same features.

28. The occupant classification system in claim 26, wherein no said expert system classifier uses the same feature as any other said expert system classifier.

29. The occupant classification system in claim 26, wherein a weighted voting scheme is used to generate a single overall occupant-type classification and a single overall confidence factor relating to said overall occupant-type classification.

30. The occupant classification system in claim 17, wherein reference classifiers are used to grade the usefulness of each said feature in said expert system classifier.

31. The occupant classification system in claim 30, wherein all features have been evaluated for effectiveness in generating accurate occupant-type classifications, and only said features rated average or higher are included in said expert system classifier.

32. The occupant classification system of claim 17, wherein there is only one said sensor.

33. The occupant classification system of claim 32, wherein said sensor is a video camera.

34. The occupant classification system of claim 17, wherein said vector of features is derived from a segmented image of the occupant.

35. The occupant classification system of claim 34, wherein said vector of features does not include a motion feature and a location feature.

36. An occupant classification system comprising:

a video camera for capturing an ambient image of a seat area and a segmented image of an occupant;

a feature extractor that receives said ambient image and said segmented image from said video camera and extracts said ambient image and said segmented image into a vector of features comprising at least one feature;

an expert system classifier that receives said vector of features and uses internal rules embodied in a decision tree to classify the occupant of said seat area into one of a plurality of predefined occupant-type classifications comprising adult, child, and non-adult seat; and

a confidence factor extractor that receives said occupant-type classification and said vector of features and determines the confidence level for the classification of said occupant.

37. The occupant classification system in claim 36, wherein more than one feature is used by said expert system classifier and more than one said expert system classifier is used by said occupant classification system, with each said expert system classifier contributing to a single overall occupant-type classification determination, and wherein each said expert system classifier is accompanied by one said confidence factor extractor, wherein each said confidence factor extractor contributes to a single overall confidence factor relating to said single overall occupant-type classification determination.

38. The occupant classification system of claim 36, said system comprising one said video camera.

39. The occupant classification system of claim 36, wherein said video camera does not capture an infrared image of said occupant.

40. The occupant classification system of claims 36, wherein said vector of features does not include a motion feature and a location feature.

41. A method for associating a predetermined occupant-type classification to an occupant image, comprising the steps of:

applying a decision tree utilizing different features and boundary values for those features leading to occupant-type classification determinations, wherein all said features are derived from the occupant image; and

extracting a confidence factor based on past performance of the decision tree.

42. The method in claim 41, comprising the additional prior step of including only those features which have been tested with regard to their utility in accurately predicting occupant type classifications.

43. The method in claim 41, comprising the step of utilizing multiple decision trees and generating one weighted occupant-type classification and one overall confidence factor.

44. The method in claim 43, comprising the step of utilizing one overall weighted confidence factor.

45. The method of claim 41, further comprising:

capturing visual images of the occupant with a video camera.

46. The method of claim 45, wherein there is only one video camera.

47. The method of claim 41, wherein said vector of features does not include an occupant location and an occupant velocity.

48. The method of claim 41, further comprising:

conducting a test to determine which features should be included in the vector of features.