Real-time embedded vision-based human hand detection

• US 8,525,876 B2
• Filed: 05/12/2010
• Issued: 09/03/2013
• Est. Priority Date: 05/12/2010
• Status: Expired due to Fees

• Current Assignee: VisionBrite Technologies Inc.
• Original Assignee: VisionBrite Technologies Inc.
• Inventors: Fan, Wensheng, Tang, WeiYi
• Primary Examiner(s): Kelley, Christopher S
• Assistant Examiner(s): Kim, Hee-Yong
• Application Number:
  US12/778,341
• Publication Number:
  US 20110279663A1
• Time in Patent Office: 1,210 Days
• Field of Search: 348/77, 345/173
• US Class Current: 348/77
• CPC Class Codes: G06V 40/28 Recognition of hand or arm movements, e.g. recognition of deaf sign language static hand signs G06V40/113
• CPC Class Codes: H04N 7/183 for receiving images from a single remote source

Abstract

In one aspect there is provided an embodiment of an image capture device comprising a camera, an image processor, a storage device and an interface. The camera is configured to capture images in ambient light of a human hand in a field of view (FOV) of the camera. The image processor is configured to process a first one of the images to detect a presence of the hand. The image capture device is configured to assign a position of the presence of the hand, track movement of the hand within the FOV by processing at least a second one of the images and generate a command based on the tracked movement of the hand within the FOV. The interface is configured to transmit the detection of the presence of the hand, the assigned position of the hand and the command to an external apparatus.

Claims

1. A method, comprising:

capturing images, with a camera of an image capture device, in ambient light of a human hand in a field of view (FOV) of said camera;

processing, by an image processor of said image capture device, a first one of said images to detect a presence of said hand, wherein said processing includes the steps of;

determining if a first contour line starting from a border of said FOV is longer than a first threshold;

determining, when said first contour line is longer than said first threshold, second contour lines for each of two edges of at least three fingers from said first one of said images of said hand in said FOV;

generating single pixel width contour lines from each of said second contour lines; and

determining if said single pixel width contour lines are finger edge or finger tip points, wherein said determining if said single pixel width contour lines are finger tip points comprises the steps of;

computing a first and second derivative of each of said single pixel width contour lines when a straight line approximation of said single pixel width contour line is greater than a second threshold; and

determining if each of said single pixel width contour lines with said straight line approximation greater than said second threshold is said finger tip point when a sign of said computed second derivative changes once;

assigning, by said image capture device, a position of said presence of said hand;

tracking, by said image capture device, movement of said hand within said FOV by processing at least a second one of said images;

generating, by said image capture device, a command based on said tracked movement of said hand within said FOV; and

transmitting, with an interface, said detection of said hand, said position of said hand, and said command to an external apparatus.

2. The method as recited in claim 1 wherein said determining if said single pixel width contour lines are finger edge lines comprises the steps of:

approximating each of said single pixel width contour lines as a straight line when said straight line approximation is below said second threshold;

determining a length of each of said approximated straight lines;

determining if each of said approximated straight lines is one of said finger edge lines when said length is greater than a third threshold; and

storing a slope and position of each of said finger edge lines in a storage device of said image capture device.

3. The method as recited in claim 2 wherein said determining if said single pixel width contour lines are finger tip points further comprises the step of:

storing a position of each of said finger tip points in said storage device of said image capture device.

4. The method as recited in claim 3 wherein said detection of said presence of said hand comprises the steps of:

determining if said stored slope of at least five finger edge lines are substantially the same; and

determining if said position of at least one of said finger tip points is between two adjacent ones of said finger edge lines.

5. The method as recited in claim 3 wherein said generating a position of said presence of said hand is based on said position of each of said finger edge lines and said position of each of said finger tip points after said hand has remained substantially still for at least 0.5 seconds.

6. The method as recited in claim 3 wherein said tracking comprises the steps of:

comparing said position for any of said stored finger edge lines in said first one of said images with a position for a same one of at least five finger edge lines determined in said at least second one of said images; and

generating said tracked movement command based on said comparing.

7. The method as recited in claim 1 wherein a relative angle of hand orientation in said FOV is not required.

8. The method as recited in claim 1 wherein said detection of said presence of said hand does not require a pre-detection training sequence with said hand.

9. The method as recited in claim 1 further comprising associating, by said external apparatus, said position of said presence of said hand with an object displayed by said external apparatus in said FOV.

10. The method as recited in claim 9 wherein said object displayed by said external apparatus in said FOV is moved corresponding to said command.

11. An image capture device, comprising:

a camera;

an image processor;

a storage device; and

an interface wherein;

said camera is configured to capture images in ambient light of a human hand in a field of view (FOV) of said camera,

said image processor is configured to;

process a first one of said images to detect a presence of said hand,

determine if a first contour line starting from a border of said FOV is longer than a first threshold,

determine, when said first contour line is longer than said first threshold, second contour lines for each of two edges of at least three fingers from said first one of said images of said hand in said FOV,

generate single pixel width contour lines from each of said second contour lines; and

determine if said single pixel width contour lines are finger edge lines or finger tip points, wherein said determining if said single pixel width contour lines are finger tip points by;

computing a first and second derivative of each of said single width contour lines when a straight line approximation of said single pixel width contour line is greater than a second threshold, and

determining if each of said single pixel width contour lines with said straight line approximation greater than said second threshold is said finger tip point when a sign of said computed second derivative changes once,

said image capture device is configured to;

assign a position of said presence of said hand,

track movement of said hand within said FOV by processing at least a second one of said images, and

generate a command based on said tracked movement of said hand within said FOV, and

said interface is configured to transmit said detection of said hand, said position of said hand, and said command to an external apparatus.

12. The image capture device as recited in claim 11 wherein said image processor is further configured to determine if said single pixel width contour lines are finger edge lines by:

approximating each of said single pixel width contour lines as a straight line when said straight line approximation is below said second threshold;

determining a length of each of said approximated straight lines; and

determining if each of said approximated straight lines is one of said finger edge lines when said length is greater than a third threshold, wherein a slope and position of each of said finger edge lines is stored in said storage device.

13. The image capture device as recited in claim 12 wherein a position of each of said finger edge lines is stored in said storage device.

14. The image capture device as recited in claim 13 wherein said image processor if further configured to detect said presence of said hand by:

determining if said stored slope of at least five finger edge lines are substantially the same; and

determining if said position of at least one of said finger tip points is between two adjacent finger edge lines.

15. The image capture device as recited in claim 13 wherein said image capture device is further configured to assign a position of said presence of said hand based on said position of each of said finger edge lines and said position of each of said finger tip points after said hand has remained substantially still for at least 0.5 seconds.

16. The image capture device as recited in claim 13 wherein said image capture device is further configured to track movement of said hand by:

comparing said position of any of said stored finger edge lines in said first one of said images with a position for a same one of at least five finger edges determined in said at least second one of said images; and

generating said tracked movement command based on said comparing.

17. The image capture device as recited in claim 11 wherein a relative angle of hand orientation in said FOV is not required.

18. The image capture device as recited in claim 11 wherein said detection of said presence of said hand does not require a pre-detection training sequence with said hand.

19. The image capture device as recited in claim 11 wherein said external apparatus is further configured to associate said position of said hand with an object displayed by said external apparatus in said FOV.

20. The image capture device as recited in claim 19 wherein said object displayed by said external apparatus in said FOV is moved corresponding to said command.