METHODS INVOLVING FACE MODEL FITTING

• US 20090257625A1
• Filed: 04/10/2008
• Published: 10/15/2009
• Est. Priority Date: 04/10/2008
• Status: Active Grant

• Current Assignee: UTC Fire & Security Americas Corporation, Inc. (Carrier Global Corporation)
• Original Assignee: General Electric Company
• Inventors: Tu, Peter Henry, Liu, Xiaoming, Wheeler, Frederick Wilson
• Granted Patent: US 8,224,037 B2
• Time in Patent Office: Days
• US Class Current: 382/118
• CPC Class Codes: G06V 10/7557 based on appearance, e.g. active appearance models [AAM]
• CPC Class Codes: G06V 40/167 using comparisons between temporally consecutive images
• CPC Class Codes: G06V 40/172 Classification, e.g. identification

Abstract

A method for face model fitting comprising, receiving a first observed image, receiving a second observed image, and fitting an active appearance model of a third image to the second observed image and the first observed image with an algorithm that includes a first function of a mean-square-error between a warped image of the second observed image and a synthesis of the active appearance model and a second function of a mean-square-error between the warped image of the second observed image and an appearance data of the first observed image.

Claims

1. A method for face model fitting comprising:

receiving a first observed image;

receiving a second observed image; and

fitting an active appearance model of a third image to the second observed image and the first observed image with an algorithm that includes a first function of a mean-square-error between a warped image of the second observed image and a synthesis of the active appearance model and a second function of a mean-square-error between the warped image of the second observed image and an appearance data of the first observed image.

2. The method of claim 1, further comprising determining whether the warped image of the second observed image and the appearance data of the first observed image correspond to the third image.

3. The method of claim 1, wherein the appearance data of the first observed image is a warped image of the first observed image.

4. The method of claim 1, wherein the first observed image and the second observed image are video frames.

5. A method for face model fitting comprising:

receiving a first observed image;

receiving a second observed image;

receiving a third observed image; and

fitting an active appearance model of a fourth image to the third observed image, the second observed image, and the first observed image with an algorithm that includes a first function of a mean-square-error between a warped image of the third observed image and a synthesis of the active appearance model and a second function of a mean-square-error between a warped image of the third observed image and appearance data of the first and second observed images.

6. The method of claim 5, further comprising determining whether the warped image of the third observed image and the appearance data of the first and second observed images correspond to the appearance model of the fourth image.

7. The method of claim 5, wherein the appearance data of the first and second observed images is a decaying average of a warped image of the first observed image and a warped image of the second observed image.

8. The method of claim 5, wherein the appearance data of the first and second observed images is an average of a warped image of the first observed image and a warped image of the second observed image.

9. The method of claim 5, wherein the appearance data of the first and second observed images is a dynamic eigenspace model of a warped image of the first observed image and a warped image of the second observed image.

10. The method of claim 5, wherein the first observed image, the second observed image, and the third observed image are video frames.

11. A method for face model fitting comprising:

receiving a first observed image;

receiving a second observed image; and

fitting an active appearance model of a third image to the second observed image and the first observed image with an algorithm

12. The method of claim 11, wherein the M_t(x) function is defined as M_t(x)=I_t-1(W(x;

p_t-1)), where t is a time associated with the second observed image and t-1 is a time associated with the first observed image.

13. The method of claim 11, wherein the method further comprises

receiving a plurality of observed images from video frames,

defining a group of L number of images, wherein the group includes the plurality of observed images and the first observed image;

and

defining the M_t(x) function as

14. The method of claim 11, wherein the method further comprises

receiving a plurality of observed images from video frames,

defining a group of L number of images, wherein the group includes the plurality of observed images and the first observed image;

and

defining the M_t(x) function as an average of warped images of the group of 1, number of images.

15. The method of claim 11,

receiving a plurality of observed images from video frames,

defining a group of L number of images, wherein the group includes the plurality of observed images and the first observed image;

and

defining the M_t(x) function as a dynamic eigenspace model of warped images of the group of L number of images.