Video occupant detection and classification

• US 5,983,147 A
• Filed: 02/06/1997
• Issued: 11/09/1999
• Est. Priority Date: 02/06/1997
• Status: Expired due to Term

• Current Assignee: Sandia Corporation (Martin Marietta Corporation)
• Original Assignee: Sandia Corporation (Martin Marietta Corporation)
• Inventors: Krumm, John C.
• Primary Examiner(s): Cuchlinski, Jr., William A.
• Assistant Examiner(s): Hernandez, Olga
• Application Number:
  US08/796,250
• Time in Patent Office: 1,006 Days
• Field of Search: 701/45, 701/28, 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: G06F 18/213 Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
• CPC Class Codes: G06T 2207/10012 Stereo images
• CPC Class Codes: G06T 2207/30268 Vehicle interior
• CPC Class Codes: G06T 7/001 using an image reference approach
• 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
• CPC Class Codes: H04N 13/218 using spatial multiplexing
• CPC Class Codes: H04N 13/239 using two 2D image sensors having a relative position equal to or related to the interocular distance H04N13/243 takes precedence
• CPC Class Codes: H04N 2013/0081 Depth or disparity estimation from stereoscopic image signals

Abstract

A system for determining when it is not safe to arm a vehicle airbag by storing representations of known situations as observed by a camera at a passenger seat; and comparing a representation of a camera output of the current situation to the stored representations to determine the known situation most closely represented by the current situation. In the preferred embodiment, the stored representations include the presence or absence of a person or infant seat in the front passenger seat of an automobile.

Claims

1. A system for distinguishing situations at a specified location comprising:

camera means comprising

a lens having a field of view encompassing the specified location, and

output means for providing an output corresponding to an image observed by said lens;

storage means for storing a vector representation of a plurality of images of a known situation as observed by a camera at the specified location, said vector representation including eigenvectors associated with each image;

signal processing means for comparing a vector representation of the camera output to the stored representations to determine the stored representation with the highest correlation to the output signal;

means for comparing the highest correlation to a predetermined value; and

indicator means for indicating the known situation of the stored signal when the highest correlation exceeds the predetermined value.

2. The system of claim 1 wherein the stored representation of known situation is represented by n'"'"' eigenvectors determined from a matrix formed from n eigenvectors representing all n output images of that situation, where n'"'"'<

<

n.

3. The system of claim 2 wherein n'"'"'=10 to 20 images.

4. The system of claim 3 wherein n'"'"'=18 images.

5. The system of claim 2 wherein the specified location is a vehicle seat, and the known situation is from the group consisting of the seat being occupied by an object or is empty.

6. The system of claim 5 further providing means for disabling a vehicle air bag opposing said seat when said seat is occupied by said object or is empty.

7. The system of claim 5 wherein the object is a child car seat.

8. The system of claim 2 wherein each vector represents the intensity of the image.

9. A system for distinguishing situations at a specified location comprising:

a camera comprising

a pair of lenses for providing binocular stereo images of the specified location, and

output means for providing an output corresponding to an image observed by said lens;

storage means for storing a vector representation of a known situation at the specified location as observed by said camera means, and

signal processing means for comparing a vector representation of the camera output to the stored representations to determine the stored representation with the highest correlation to the output signal, wherein each vector represents a function of the range of the image from said camera;

means for comparing the highest correlation to a predetermined value; and

indicator means for indicating the known situation of the stored signal when the highest correlation exceeds the predetermined value.

10. The system of claim 9 wherein each vector represents disparity of the stereo images.

11. The system of claim 9 wherein the specified location is a vehicle seat, and the known situations consist of the seat being occupied by a person, an object, being empty.

12. The system of claim 11 further providing means for disabling a vehicle air bag opposing said seat when said seat is occupied by said object or is empty.

13. The system of claim 11 wherein said signal processing means determines the size of the person from the range information, and the weight of the person from the size.

14. The system of claim 11 wherein said signal processing means determines the position of the person in the seat from the range information.

15. The system of claim 11 wherein said signal processing means determines the distance between the person in the seat from landmarks in the vehicle from the range information.

16. An efficient system for determining if an image corresponds to a predetermined situation having many variations, comprising:

means for generating a vector v_i representing said first image;

means for correlating v_i with a function of n'"'"' of n vectors p_i representing possible variations of the situation, wherein n'"'"'<

<

n, and correlation which exceeds a predetermined value provides an indication that the image corresponds to the situation.

17. The efficient system of claim 16 further comprising means for determining which n'"'"' of the n vectors shall be used, said means comprising:

means for combining the n vectors as a matrix P;

means for determining the eigenvectors of the covariance matrix of P; and

means for expressing each vector p'"'"' as a weighted sum of the n'"'"' eigenvectors associated with the highest eigenvalues.

18. An efficient method for determining if an image corresponds to a predetermined situation having many variations, comprising:

capturing an image;

generating a vector v_i representing said image;

correlating v_i with a function of n'"'"' of n vectors p_i representing possible variations of a situation, wherein n'"'"'<

<

n, and correlation which exceeds a predetermined value provides an indication that the image corresponds to the situation.

19. The efficient method of claim 18 further comprising the step of determining which n'"'"' of the n vectors shall be used by:

combining the n vectors as a matrix P;

determining the eigenvectors of the covariance matrix of P; and

expressing each vector p_i as a weighted sum of the n'"'"' eigenvectors associated with the highest eigenvalues.