METHOD FOR IDENTIFYING A PERSON'S POSTURE

• US 20090012921A1
• Filed: 02/02/2007
• Published: 01/08/2009
• Est. Priority Date: 02/02/2006
• Status: Active Grant

• Current Assignee: Commissariat a L'Energie Atomique (AES Corporation)
• Original Assignee: Commissariat a L'Energie Atomique (AES Corporation)
• Inventors: Godin, Christelle, Bonnet, Stephane
• Granted Patent: US 9,026,477 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G06V 30/242 Division of the character sequences into groups prior to recognition; Selection of dictionaries
• CPC Class Codes: G06V 40/23 Recognition of whole body movements, e.g. for sport training

Abstract

A classification method including first classifying an event of any kind by first rules, and then second classifying events, not identified by the first classification, by a learning base reinforced with all the events identified by the first classification. The method is adaptive if the second classification rules are amended according to new examples that were able to be determined by the first rules.

Claims

1-13. -13. (canceled)

14. :

A method for identifying postures of a person, intended for identifying events corresponding to a particular posture or to a change in posture and associated with respective classes, comprising;

attaching a sensor on the person;

recording at least one signal from the sensor during a time window;

extracting partitioning parameters from the signal;

dividing, from the partitioning parameters, the time window into segments meeting a homogeneity criterion of posture or change in posture;

extracting, for each of the segments, a first set of parameters from the signal;

classifying each of the segments in one of event classes associated with events or in a class that is not associated with an event, by applying first decision rules from the first set of parameters associated with the segments;

a calibration operation to create a learning base, comprising, for each segment belonging to an event class;

extracting a second set of parameters from the signal;

adding into the learning base a datum listing the second set of parameters and the event class associated with the relevant segment; and

using the learning base to classify segments of the recording or of another subsequent recording that have not been associated with an event by applying second decision rules using the learning base from a second set of parameters associated with these segments.

15. :

The method for identifying postures of a person according to claim 14, wherein the sensor comprises a triaxial accelerator and a triaxial magnetometer.

16. :

The method for identifying postures of a person according to claim 14, wherein at least some of the segments identified by the first and the second decision rules are incorporated into the learning base by their parameters and the postures associated therewith.

17. :

The method for identifying postures of a person according to claim 14, further comprising sampling the signal into portions with uniform durations and grouping groups of portions to provide segments.

18. :

The method for identifying postures of a person according to claim 17, wherein the grouping of groups of portions is carried out according to a criterion of maximum variation of an activity level of the person and a criterion of maximum tilt variation of the person.

19. :

The method for identifying postures of a person according to claim 17, further comprising distributing the segments into two alternating categories.

20. :

The method for identifying postures of a person according to claim 14, wherein the first decision rules comprise distinguishing between horizontal postures and vertical postures according to an orientation measurement of acceleration of gravity by the sensor.

21. :

The method for identifying postures of a person according to claim 20, wherein the first decision rules comprise distinguishing between a standing posture and a sitting posture according to at least duration and activity level criteria.

22. :

The method for identifying postures of a person according to claim 20, wherein the first decision rules comprise identifying transient states immediately between pairs of states of different postures separating the transient states, and eliminating transient states immediately between pairs of states with similar postures.

23. :

The method for identifying postures of a person according to claim 14, wherein the second group of parameters comprises for each of the segments, statistical parameters on the signals, duration of the segment, frequency content of the segment.

24. :

The method for identifying postures of a person according to claim 14, wherein the second group of parameters comprises a signature of segments, or at least one component of the signal reflecting a developmental function of the posture.

25. :

The method for identifying postures of a person according to claim 14, wherein the second decision rules depend on similarity scores or on distance between the segments to be identified and the segments added into the learning base according to the postures associated with the added segments.

26. :

The method for identifying postures of a person according to claim 14, wherein the sensor comprises gyrometers and the second decision rules comprise a comparison of an average azimuth in the segments with a reference azimuth.