SYSTEM AND METHOD FOR COMPUTER VISION BASED TRACKING OF AN OBJECT

• US 20150178556A1
• Filed: 01/19/2015
• Published: 06/25/2015
• Est. Priority Date: 05/10/2012
• Status: Active Grant

• Current Assignee: PointGrab Ltd.
• Original Assignee: PointGrab Ltd.
• Inventors: PERSKI, HAIM, BITTON, DAVID
• Granted Patent: US 9,256,781 B2
• Time in Patent Office: Days
• 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: G06F 3/0304 Detection arrangements using opto-electronic means constructional details of pointing devices not related to the detection arrangement using opto-electronic means G06F3/033; optical digitisers G06F3/042
• 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

A system and method for computer vision based tracking of a human form may include receiving a sequence of images, the images including at least one object having a shape of a human form. A first selected feature is tracked from within the human form shaped object. Shape recognition algorithms are applied at a suspected location of the human form shaped object in an image from the sequence of images to detect a shape of a human form in the image and a second feature from within the detected shape of the human form is then selected and tracked, thereby providing verification and updating of the location of the human form shaped object.

Claims

1. A method for computer vision based tracking of an object, the method comprising:

i) receiving a sequence of images of a space, said images comprising at least one object having a shape of a human form;

ii) tracking at least one first selected feature from within the shape of the object;

iii) applying a shape recognition algorithm at a suspected location of the object in an image from the sequence of images to detect a newly detected shape of a human form in the image;

iv) selecting at least one newly selected feature to be tracked from within the newly detected shape of the human form; and

v) tracking the newly selected feature.

2. The method of claim 1 comprising repeating steps (iii)-(v) to track a human form throughout a plurality of images.

3. The method of claim 2 comprising:

assigning a probability grade to a detected shape for each of the plurality of images;

using the probability grades to obtain a combined probability grade; and

determining that no shape of a human form is detected if the combined probability grade is below a predetermined threshold.

4. The method of claim 1 comprising terminating tracking if no shape of a human form is detected at the suspected location.

5. The method of claim 1 wherein the first and newly selected features comprise pixels, patterns or vectors.

6. The method of claim 1 wherein tracking the first feature and the newly selected feature comprises tracking a human form, the method further comprising controlling a device according to the tracking of the human form.

7. The method of claim 6 wherein controlling a device comprises turning the device ON or OFF or modulating output of the device.

8. The method of claim 1 comprising

classifying a first set of selected features to groups;

determining multiple suspected locations according to locations of the groups;

applying a shape recognition algorithm at each suspected location;

determining, based on the shape recognition algorithm, a shape at one of the suspected locations, with the highest probability of being a shape of a human form;

selecting at least one newly selected feature to be tracked from within the determined shape; and

tracking the newly selected feature.

9. The method of claim 1 wherein the images are top view images of the space.

10. The method of claim 9 wherein applying a shape recognition algorithm comprises obtaining rotation invariant descriptors from the images.

11. A method for computer vision based tracking of a human form in a sequence of images, the method comprising:

detecting in a first image an object having a shape of a human form;

tracking at least one first selected feature from within the shape of the human form;

detecting in a second image a shape of a human form by obtaining rotation invariant descriptors from the second image at a suspected location of the human form in the second image;

selecting at least one feature from within the shape of the human form detected in the second image; and

tracking the feature selected from within the shape of the human form in the second image.

12. The method of claim 11 comprising

determining occupancy in a space based on the tracking of the human form;

and

controlling a device based on the determination of occupancy.

13. The method of claim 12 wherein controlling a device comprises turning the device ON or OFF or modulating output of the device.

14. A system for computer vision based tracking of a human form, the system comprising

a processor configured to:

detect within a first image from a sequence of images an object having a shape of a human form;

track at least one feature selected from within the shape of the human form;

detect in a second image a shape of a human form at a suspected location;

select at least one feature to be tracked from within the shape of the human form detected in the second image; and

track the feature selected from within the shape of the human form detected in the second image; and

determine occupancy in a space based on the tracking of the human form.

15. The system of claim 14 comprising an image sensor to obtain the sequence of images, said image sensor in communication with the processor.

16. The system of claim 14 wherein the processor is to generate a signal to control a device based on the determination of occupancy.

17. The system of claim 16 wherein the signal is to turn the device ON or OFF or modulate output of the device.