SYSTEMS AND METHODS FOR GENERATING INDIVIDUALIZED 3D HEAD MODELS

• US 20090153552A1
• Filed: 11/19/2008
• Published: 06/18/2009
• Est. Priority Date: 11/20/2007
• Status: Abandoned Application

• Current Assignee: Image Metrics, Inc.
• Original Assignee: Big Stage Entertainment, Inc. (Image Metrics, Inc.)
• Inventors: Mo, Zhenyao, Mediomi, Gerard Guy, Lewis, John Peter, Johnson, Michael David, Fidaleo, Douglas Alexander
• Application Number:
  US12/274,226
• Publication Number:
  US 20090153552A1
• Time in Patent Office: Days
• US Class Current: 345/419
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0247 Calculate past, present or future revenues
• CPC Class Codes: G06T 13/40 of characters, e.g. humans, animals or virtual beings
• CPC Class Codes: G06T 15/00 3D [Three Dimensional] image rendering
• CPC Class Codes: G06T 19/20 Editing of 3D images, e.g. changing shapes or colours, aligning objects or positioning parts
• CPC Class Codes: G06T 2219/2004 Aligning objects, relative positioning of parts
• CPC Class Codes: G06T 2219/2021 Shape modification

Abstract

Systems and methods are disclosed for generating a substantially life-like, individualized three-dimensional (3D) head model. In one example, the method includes reconstructing a 3D face mask from the input image data of an individual'"'"'s head or face. Detected facial landmarks of the reconstructed face mask are aligned with corresponding facial landmarks of a generic head model, and the remainder of the head is reconstructed by generating a blend fit head model from a weighted combination of basis head models. The blend fit model is then deformed to more accurately reflect the facial and/or head geometry of the individual. The method may further include associating an animation rig with the individualized 3D head model that allows for simulated movement of the 3D head model in, for example, personalized media content.

Claims

1. A method for computer modeling a human head, the method comprising:

receiving a plurality of input images comprising a human head, each of the plurality of input images being associated with a different view of the human head;

reconstructing a textured three-dimensional (3D) face mask from the plurality of input images, the 3D face mask having a non-uniform surface;

detecting a plurality of first landmark locations of the 3D face mask, each of the plurality of first landmark locations corresponding to selected facial locations of the 3D face mask;

aligning the plurality of first landmark locations of the 3D face mask with a plurality of second landmark locations corresponding to a generic head model;

comparing the aligned 3D face mask with a plurality of basis head models, each of the plurality of basis head models comprising a unique set of facial features;

generating a blend fit head model from a combination of selected ones of the plurality of basis head models, the combination of the selected basis head models approximating parameters of the aligned 3D face mask;

deforming portions of the blend fit head model to approximate the non-uniform surface of the aligned 3D face mask to generate a personalized 3D head model; and

storing the personalized 3D head model in a memory.

2. The method of claim 1, wherein the combination of the selected basis head models comprises a weighted combination.

3. The method of claim 1, wherein the plurality of input images comprises at least two still images obtained from a digital camera.

4. The method of claim 1, additionally comprising:

extracting at least one color value at one or more pixel locations proximate a perimeter of a face portion of the personalized 3D head model; and

extrapolating the at least one color value to a remainder of the personalized 3D head model outside the face portion.

5. The method of claim 1, additionally comprising associating an animation rig with the personalized 3D head model, the animation rig defining a deformation of the personalized 3D head model to simulate movement of the personalized 3D head model.

6. The method of claim 1, wherein the plurality of first landmark locations comprises at least a nose tip, mouth corners, and eye corners.

7. The method of claim 1, wherein said generating the blend fit model comprises performing a least squares fitting comparison between the plurality of basis head models and the aligned 3D face mask.

8. The method of claim 1, wherein said aligning comprises performing an iterative closest point comparison between the plurality of first landmark locations and the plurality of second landmark locations.

9. The method of claim 1, wherein said deforming comprises a localized, unconstrained process of deforming the portions of the blend fit head model.

10. The method of claim 9, wherein said deforming further comprises performing a radial-basis function interpolation on portions of the blend fit head model to align facial features of the blend fit head model with corresponding facial features of the 3D face mask.

11. A system for generating a model of a head, the system comprising:

head data representing at least one image comprising at least a frontal portion of a face, the head data further comprising first landmark data identifying a plurality of first facial locations of the face;

generic head model data representing a three-dimensional (3D) generic head model, the generic head model data further comprising second landmark data identifying a plurality of second facial locations of the 3D generic head model;

basis head model data representing a plurality of 3D basis head models, each of the 3D basis head models comprising a unique set of facial attributes associated therewith; and

a processor configured to receive the head data, the generic head model data and the basis head model data to,

align the at least one image with the 3D generic head model by comparing the first landmark data with the second landmark data,

compare the aligned at least one image with the plurality of 3D basis head models,

create, based on said comparison, a 3D blend fit head model from a combination of selected ones of the plurality of 3D basis head models, and

deform portions of the 3D blend fit head model to approximate a non-uniform surface of the aligned at least one image to generate a personalized 3D head model.

12. The system of claim 11, wherein the at least one image comprises a 3D face mask, wherein the system further comprises a mask module configured to generate the head data from input image data.

13. The system of claim 12, wherein the input image data comprises a plurality of images, each of the plurality of images being associated with a different view of the head.

14. The system of claim 11, wherein the unique set of attributes for each 3D basis head model indicates surface contours of the respective 3D basis head model.

15. The system of claim 11, further comprising an extrapolation module configured to extrapolate one or more color values at one or more pixel locations proximate a perimeter of a face portion of the personalized 3D head model to a remainder of the personalized 3D head model outside the face portion.

16. The system of claim 11, wherein the processor is further configured to associate animation data with the personalized 3D head model.

17. The system of claim 16, wherein the animation data defines one or more deformation processes of the personalized 3D head model to simulate movement of the personalized 3D head model.

18. The system of claim 11, further comprising a memory configured to store the generic head model data and the basis head model data.

19. A system for modeling a human head, the system comprising:

means for receiving a plurality of first landmark locations of at least one head image of at least a frontal portion of a face, the at least one head image having a non-uniform surface;

means for aligning the plurality of first landmark locations of the at least one head image with a plurality of second landmark locations corresponding to a generic head model;

means for comparing the aligned 3D at least one head image with a plurality of basis head models, each of the plurality of basis head models comprising a unique set of facial features;

means for generating a blend fit head model from a weighted combination of selected ones of the plurality of basis head models, the weighted combination of the selected basis head models approximating parameters of the aligned at least one head image; and

means for deforming portions of the blend fit head model to approximate the non-uniform surface of the aligned at least one head image to generate a personalized 3D head model.

20. The system of claim 19, further comprising means for extrapolating one or more skin colors at one or more pixel locations proximate a perimeter of a face region of the personalized 3D head model to a remainder of the personalized 3D head model outside the face region.