Interpreting Sign Language Gestures

• US 20090040215A1
• Filed: 08/10/2007
• Published: 02/12/2009
• Est. Priority Date: 08/10/2007
• Status: Active Grant

• Current Assignee: Asian Institute of Technology
• Original Assignee: Asian Institute of Technology
• Inventors: Phuc, Trinh Huu, Afzulpurkar, Nitin
• Granted Patent: US 7,949,157 B2
• Time in Patent Office: Days
• US Class Current: 345/419
• 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

Embodiments of the invention include methods and apparatuses relating to identifying hand gestures. In one embodiment, a hand gesture is identified by capturing an image including an image object of the hand gesture, generating a numerical signature based on the image object, and identifying the hand gesture from reference numerical signatures associated with reference hand gestures.

Claims

1. A hand gesture recognition method comprising:

obtaining a captured image including an image object, the image object being an image of a hand;

generating a numerical signature based on the image object;

providing a plurality of reference numerical signatures, the plurality of reference numerical signatures having a plurality of reference hand gestures associated therewith; and

identifying one of the plurality of reference numerical signatures as being descriptive of the numerical signature, wherein a hand gesture corresponding to the one of the plurality of reference numerical signatures is attributable to the image of the hand.

2. The hand gesture recognition method of claim 1, wherein the numerical signature comprises a dimensional ratio of the image object, the dimensional ratio being calculated as:

$d=\frac{W}{H},$

wherein d is the dimensional ratio, W is a width of the image object, and H is a height of the image object.

3. The hand gesture recognition method of claim 1, wherein the numerical signature comprises a dimensional ratio of the image object, the dimensional ratio being calculated as:

$d=\frac{\mathrm{BCV}}{H},$

wherein d is the dimensional ratio, BCV is the length of a segment created by an intersection between a boundary of the image object and a vertical scanning line, and H is a height of the image object.

4. The hand gesture recognition method of claim 3, wherein the vertical scanning line is scanned at about 34% of the width of the image object.

5. The hand gesture recognition method of claim 1, wherein the numerical signature comprises a number of intersections between a boundary of the image object and a horizontal scanning line.

6. The hand gesture recognition method of claim 1, wherein the hand gesture comprises one of the American Sign Language (ASL) gestures.

7. The hand gesture recognition method of claim 1, further comprising:

arranging the plurality of reference hand gestures in a hierarchical structure according to the plurality of reference numerical signatures corresponding thereto, wherein the hierarchical structure enables searching of the plurality of reference hand gestures.

8. The hand gesture recognition method of claim 1, further comprising:

obtaining a second captured image including a second image object, the second image object being a second image of the hand;

determining whether the second captured image is substantially similar to the captured image; and

if the second captured image and the captured image are not substantially similar, attributing a second hand gesture to the captured image and the second captured image.

9. A machine readable medium having stored therein a plurality of instructions that, when executed, cause the machine to identify a hand gesture by:

receiving a captured image including an image object, the image object being an image of a hand;

generating a numerical signature based on the image object;

receiving a plurality of reference numerical signatures, the plurality of reference numerical signatures having a plurality of reference hand gestures associated therewith; and

identifying one of the plurality of reference numerical signatures being descriptive of the numerical signature, wherein a hand gesture corresponding to the one of the plurality of reference numerical signatures is attributable to the image of the hand.

10. The machine readable medium of claim 9, wherein the numerical signature comprises a dimensional ratio of the image object, the dimensional ratio being calculated as:

$d=\frac{W}{H},$

wherein d is the dimensional ratio, W is a width of the image object, and H is a height of the image object.

11. The machine readable medium of claim 9, wherein the numerical signature comprises a dimensional ratio of the image object, the dimensional ratio being calculated as:

$d=\frac{\mathrm{BCV}}{H},$

wherein d is the dimensional ratio, BCV is the length of a segment created by an intersection between a boundary of the image object and a vertical scanning line, and H is a height of the image object.

12. The machine readable medium of claim 11, wherein the vertical scanning line is scanned at about 34% of the width of the image object.

13. The machine readable medium of claim 9, wherein the numerical signature comprises a number of intersections between a boundary of the image object and a horizontal scanning line.

14. The machine readable medium of claim 9, wherein the hand gesture comprises one of the American Sign Language (ASL) gestures.

15. The machine readable medium of claim 9, further comprising instructions that, when executed, cause the machine to identify a hand gesture by:

arranging the plurality of reference hand gestures in a hierarchical structure according to the plurality of reference numerical signatures corresponding thereto, wherein the hierarchical structure enables searching of the plurality of reference hand gestures.

16. The machine readable medium of claim 9, further comprising instructions that, when executed, cause the machine to identify a hand gesture by:

obtaining a second captured image including a second image object, the second image object being a second image of the hand;

determining whether the second captured image is substantially similar to the captured image; and

if the second captured image and the captured image are not substantially similar, attributing a second hand gesture to the captured image and the second captured image.

17. An apparatus comprising:

a machine readable medium having stored therein a plurality of instructions that, when executed, cause the apparatus to identify a hand gesture by;

receiving a captured image including an image object, the image object being an image of a hand;

generating a numerical signature based on the image object;

receiving a plurality of reference numerical signatures, the plurality of reference numerical signatures having a plurality of reference hand gestures associated therewith; and

identifying one of the plurality of reference numerical signatures being descriptive of the numerical signature, wherein a hand gesture corresponding to the one of the plurality of reference numerical signatures is attributable to the image of the hand; and

a processor coupled to the machine readable medium to execute the programming instructions.

18. The apparatus of claim 17, wherein the numerical signature comprises a dimensional ratio of the image object, the dimensional ratio being calculated as:

$d=\frac{W}{H},$

wherein d is the dimensional ratio, W is a width of the image object, and H is a height of the image object.

19. The apparatus of claim 17, wherein the numerical signature comprises a dimensional ratio of the image object, the dimensional ratio being calculated as:

$d=\frac{\mathrm{BCV}}{H},$

wherein d is the dimensional ratio, BCV is the length of a segment created by an intersection between a boundary of the image object and a vertical scanning line, and H is a height of the image object.

20. The apparatus of claim 19, wherein the vertical scanning line is scanned at about 34% of the width of the image object.

21. The apparatus of claim 17, wherein the numerical signature comprises a number of intersections between a boundary of the image object and a horizontal scanning line.

22. The apparatus of claim 17, wherein the hand gesture comprises one of American Sign Language (ASL) gestures.

23. The apparatus of claim 17, wherein the machine readable medium further comprises instructions that, when executed, cause the apparatus to identify a hand gesture by:

arranging the plurality of reference hand gestures in a hierarchical structure according to the plurality of reference numerical signatures corresponding thereto, wherein the hierarchical structure enables searching of the plurality of reference hand gestures.

24. The apparatus of claim 17, wherein the machine readable machine readable medium further comprises instructions that, when executed, cause the apparatus to identify a hand gesture by:

obtaining a second captured image including a second image object, the second image object being a second image of the hand;

determining whether the second captured image is substantially similar to the captured image; and

if the second captured image and the captured image are not substantially similar, attributing a second hand gesture to the captured image and the second captured image.