Translation and scale invariant features for gesture recognition

• US 9,405,375 B2
• Filed: 09/13/2013
• Issued: 08/02/2016
• Est. Priority Date: 09/13/2013
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Mittal, Arpit
• Primary Examiner(s): Tsai, Tsung-Yin
• Application Number:
  US14/026,973
• Publication Number:
  US 20150077322A1
• Time in Patent Office: 1,054 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 3/005 Input arrangements through a video camera
• CPC Class Codes: G06F 3/017 Gesture based interaction, e.g. based on a set of recognized hand gestures interaction based on gestures traced on a digitiser G06F3/04883
• CPC Class Codes: G06V 40/28 Recognition of hand or arm movements, e.g. recognition of deaf sign language static hand signs G06V40/113

Abstract

Methods and apparatuses of the present disclosure are presented for recognizing a gesture of a gesture object in a plurality of recorded data objects, with the recorded data objects being recorded over time. In some embodiments, a method includes computing at least one set of gesture angles using the plurality of recorded data objects, wherein each of the gesture angles in the at least one set comprises an angle measurement between two positions of the gesture object, the two positions recorded in successive data objects in the plurality of recorded data objects, and recognizing the gesture based on the at least one set of gesture angles. In some embodiments, the method includes recognizing the gesture is based further on comparing the at least one set of gesture angles to a gesture model.

Claims

1. A method for recognizing a gesture of a gesture object in a plurality of recorded data objects, the recorded data objects being recorded over time, the method comprising:

determining at least one set of gesture angles using the plurality of recorded data objects, wherein each of the gesture angles in the at least one set of gesture angles comprises an angle measurement between two positions of the gesture object, the two positions recorded in successive data objects of the plurality of recorded data objects, wherein the at least one set of gesture angles further comprises a first subset of gesture angles and a second subset of gesture angles;

determining a first histogram representing a frequency of angles based on the first subset of gesture angles and a second histogram representing a frequency of angles based on the second subset of gesture angles;

recognizing the gesture based on a comparison of the first histogram and the second histogram to a respective first model histogram and second model histogram, each model histogram representing a frequency of angles of a subdivision of gestures of a gesture model; and

modifying a behavior of the device in response to the recognizing the gesture.

2. The method of claim 1, wherein the first subset of gesture angles and the second subset of gesture angles are time-ordered;

and

the recognizing the gesture includes comparing the first subset of gesture angles and the second subset of gesture angles to a respective subdivision of the gesture model.

3. The method of claim 1, wherein the first subset of gesture angles shares gesture angles with the second subset of gesture angles.

4. The method of claim 1, wherein the at least one of gesture angles further comprises a third subset of gesture angles, and wherein the third subset of gesture angles comprises a subdivision of the second subset of gesture angles.

5. The method of claim 4, wherein the recognizing the gesture is based further on comparing a third histogram representing a frequency of angles based on the third subset of gesture angles to a third model histogram representing a frequency of angles of a subdivision of gesture angles represented by the second model histogram.

6. The method of claim 1, wherein the at least one subset of gesture angles further comprises a third subset of gesture angles, and wherein the third subset of gesture angles comprises a subdivision of the first subset of gesture angles that does not include any of the gesture angles comprising the second subset of gesture angles.

7. The method of claim 6, wherein the recognizing the gesture is based further on comparing a third histogram representing a frequency of angles based on the third subset of gesture angles to a third model histogram representing a frequency of angles of a subdivision of gestures of the gesture model that does not include any gestures of the second model histogram.

8. The method of claim 7, wherein the recognizing the gesture is based further on:

concatenating the second subset of gesture angles with the third subset of gesture angles to generate a concatenated subset of gesture angles;

combining the second model histogram and the third model histogram to generate a concatenated model histogram;

and

comparing the concatenated subset of gesture angles with the concatenated model histogram.

9. The method of claim 1, wherein the gesture model comprises at least one trained set of gesture angles that was generated prior to recognizing the gesture of the gesture object.

10. An apparatus comprising:

at least one recording device configured to record a gesture object in a plurality of data objects over time; and

a processor configured to;

determine at least one set of gesture angles using the plurality of recorded data objects, wherein each of the gesture angles in the at least one set of gesture angles comprises an angle measurement between two positions of the gesture object, the two positions recorded in successive data objects of the plurality of recorded data objects, wherein the at least one set of gesture angles further comprises a first subset of gesture angles and a second subset of gesture angles;

determine a first histogram representing a frequency of angles based on the first subset of gesture angles and a second histogram representing a frequency of angles based on the second subset of gesture angles;

recognize a gesture based on comparing the first histogram and the second histogram to a respective first model histogram and second model histogram, each model histogram representing a frequency of angles of a subdivision of gestures of a gesture model; and

modify a behavior of the device in response to the recognizing the gesture.

11. The apparatus of claim 10, wherein the first subset of gesture angles and the second subset of gesture angles are time-ordered;

and

the processor is further configured to recognize the gesture including comparing the first subset of gesture angles and the second subset of gesture angles to a respective subdivision of the gesture model.

12. The apparatus of claim 10, wherein the first subset of gesture angles shares gesture angles with the second subset of gesture angles.

13. The apparatus of claim 10, wherein the at least one set of gesture angles further comprises a third subset of gesture angles, and wherein the third subset of gesture angles comprises a subdivision of the second subset of gesture angles.

14. The apparatus of claim 13, wherein the processor is further configured to compare a third histogram representing a frequency of angles based on the third subset of gesture angles to a third model histogram representing a frequency of angles of a subdivision of gesture angles represented by the second model histogram.

15. The apparatus of claim 10, wherein the at least one subset of gesture angles further comprises a third subset of gesture angles, and wherein the third subset of gesture angles comprises a subdivision of the first subset of gesture angles that does not include any of the gesture angles comprising the second subset of gesture angles.

16. The apparatus of claim 15, wherein the processor is further configured to compare a third histogram representing a frequency of angles based on the third subset of gesture angles to a third model histogram representing a frequency of angles of a subdivision of gestures of the gesture model that does not include any gestures of the second model histogram.

17. The apparatus of claim 16, wherein the processor is further configured to:

concatenate the second subset of gesture angles with the third subset of gesture angles to generate a concatenated subset of gesture angles;

combine the second model histogram and the third model histogram to generate a concatenated model histogram;

and

compare the concatenated subset of gesture angles with the concatenated model histogram.

18. The apparatus of claim 10, wherein the gesture model comprises at least one trained set of gesture angles that was generated prior to recognizing the gesture of the gesture object.

19. An apparatus for recognizing a gesture of a gesture object in a plurality of recorded data objects, the recorded data objects being recorded over time, the apparatus comprising:

means for determining at least one set of gesture angles using the plurality of recorded data objects, wherein each of the gesture angles in the at least one set of gesture angles comprises an angle measurement between two positions of the gesture object, the two positions recorded in successive data objects of the plurality of recorded data objects, wherein the at least one set of gesture angles further comprises a first subset of gesture angles and a second subset of gesture angles;

means for determining a first histogram representing a frequency of angles based on the first subset of gesture angles and a second histogram representing a frequency of angles based on the second subset of gesture angles;

means for recognizing the gesture based on comparing the first histogram and the second histogram to a respective first model histogram and second model histogram, each model histogram representing a frequency of angles of a subdivision of gestures of a gesture model; and

means for modifying a behavior of the device in response to the recognizing the gesture.

20. The apparatus of claim 19, wherein the first subset of gesture angles and the second subset of gesture angles are time-ordered;

and

the means for recognizing the gesture includes comparing the first subset of gesture angles and the second subset of gesture angles to a respective subdivision of the gesture model.

21. The apparatus of claim 19, wherein the first subset of gesture angles shares gesture angles with the second subset of gesture angles.

22. The apparatus of claim 21, wherein the at least one set of gesture angles further comprises a third subset of gesture angles, and wherein the third subset of gesture angles comprises a subdivision of the second subset of gesture angles.

23. The apparatus of claim 22, wherein the means for recognizing the gesture are based further on comparing a third histogram representing a frequency of angles based on the third subset of gesture angles to a third model histogram representing a frequency of angles of a subdivision of gesture angles represented by the second model histogram.

24. The apparatus of claim 19, wherein the at least one subset of gesture angles further comprises a third subset of gesture angles, and wherein the third subset of gesture angles comprises a subdivision of the first subset of gesture angles that does not include any of the gesture angles comprising the second subset of gesture angles.

25. The apparatus of claim 24, wherein the means for recognizing the gesture is based further on means for comparing a third histogram representing a frequency of angles based on the third subset of gesture angles to a third model histogram representing a frequency of angles of a subdivision of gestures of the gesture model that does not include any gestures of the second model histogram.

26. The apparatus of claim 25, wherein the means for recognizing the gesture is based further on:

means for concatenating the second subset of gesture angles with the third subset of gesture angles to generate a concatenated subset of gesture angles;

means for combining the second model histogram and the third model histogram to generate a concatenated model histogram;

and

means for comparing the concatenated subset of gesture angles with the concatenated model histogram.

27. The apparatus of claim 19, wherein the gesture model comprises at least one trained set of gesture angles that was generated prior to recognizing the gesture of the gesture object.

28. A non-transitory processor-readable medium comprising processor-readable instructions configured to cause a processor to:

record a gesture object in a plurality of data objects over time;

determine at least one set of gesture angles using the plurality of recorded data objects, wherein each of the gesture angles in the at least one set of gesture angles comprises an angle measurement between two positions of the gesture object, the two positions recorded in successive data objects of the plurality of recorded data objects, wherein the at least one set of gesture angles further comprises a first subset of gesture angles and a second subset of gesture angles;

determine a first histogram representing a frequency of angles based on the first subset of gesture angles and a second histogram representing a frequency of angles based on the second subset of gesture angles;

recognize a gesture based on comparing the first histogram and the second histogram to a respective first model histogram and second model histogram, each model histogram representing a frequency of angles of a subdivision of gestures of a gesture model; and

modify a behavior of the device in response to the recognizing the gesture.

29. The non-transitory processor-readable medium of claim 28, wherein the first subset of gesture angles and the second subset of gesture angles are time-ordered;

and

the instructions are further configured to cause the processor to recognize the gesture including individually comparing the first subset of gesture angles and the second subset of gesture angles to a respective subdivision of the gesture model.

30. The non-transitory processor-readable medium of claim 28, wherein the first subset of gesture angles shares gesture angles with the second subset of gesture angles.