Three-dimensional face identification system

• US 6,381,346 B1
• Filed: 12/01/1998
• Issued: 04/30/2002
• Est. Priority Date: 12/01/1997
• Status: Expired due to Fees

• Current Assignee: Wheeling Jesuit University
• Original Assignee: Wheeling Jesuit University
• Inventors: Eraslan, Arsev H.
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): CHAWAN, SHEELA C
• Application Number:
  US09/203,683
• Time in Patent Office: 1,246 Days
• Field of Search: 382/118, 382/190, 382/154, 382/100, 382/254, 382/199, 382/284, 382/285, 345/420, 345/49
• US Class Current: 382/118
• CPC Class Codes: G06T 11/00 2D [Two Dimensional] image generation
• CPC Class Codes: G06T 17/00 Three dimensional [3D] modelling, e.g. data description of 3D objects
• CPC Class Codes: G06T 17/20 Finite element generation, e.g. wire-frame surface description, tesselation
• CPC Class Codes: G06V 40/16 Human faces, e.g. facial parts, sketches or expressions

Abstract

A three-dimensional (3-D) facial imaging system is disclosed for generating facial images, indexing those images by composite codes and for searching for similar two-dimensional (2-D) facial images. The 3-D images of human faces are generated from a data repository of 3-D facial feature surface shapes. These shapes are organized by facial feature parts. By assembling a shape for each facial part is assembled, a 3-D facial image is formed. Human facial images may be represented with a composite code (facial feature part code and shape code). A 3-D image of any face may be generated using the code to select the proper facial feature shapes from the stored repository of shapes.

Claims

1. A method for generating an image using a memory unit, a processor and a display, wherein the method comprises the following steps:

(a) storing in the memory unit image feature part shapes, wherein the stored image feature part shapes are three-dimensional surface images of image parts;

(b) assigning a feature code to each image feature part shape;

(c) analyzing with the processor a first image to determine the image feature part shapes selected from memory that best match the first image;

(d) identifying the feature codes for the image features part shapes that match the first image;

(e) storing in memory the feature codes for the first image that was identified in step (d);

(f) generating, using the processor, a second image from a three-dimensional assembly of the image feature part shapes stored in step (a) that correspond to the feature codes identified for the first image in step (d);

(g) showing the second image generated in step (f) on the display;

(h) storing in a memory unit a listing of image codes comprised of a set of feature codes where each image code corresponds to a particular image;

(i) matching the feature codes identified in step (d) to at least one of the image codes stored in step (h);

(j) generating a third image corresponding to the at least one image code matched in step (i) from an assembly of the image feature shapes, and

(k) comparing the third image generated in step (j) to the second image generated in step (f).

2. A method as in claim 1 wherein the first image and the second image appear to be images of the same object.

3. A method as in claim 2 wherein the object is a human face.

4. A method as in claim 3 wherein the image codes in step (h) each correspond to a particular human face.

5. A method as in claim 4 wherein said image feature part shapes include surface shapes of a plurality of shapes of each of a nose, eye, ear and chin.

6. A method as in claim 1 wherein the first image is at least one two-dimensional image.

7. A method as in claim 1 wherein the first image comprises a first two-dimensional image, and a second two-dimensional image orthogonal to the first two-dimensional image.

8. A method as in claim 1 wherein the third image is a three-dimensional image.

9. A method as in claim 1 wherein steps (h), (i), (j) and (k) are performed on a computer remote from the computer used to perform steps (c), (d), (e), (f) and (g).

10. A method for recognizing a facial image using at least one computer, said method comprising the steps of:

(a) storing a three-dimensional images of facial part images and a facial feature code assigned to each facial part image;

(b) analyzing a first facial image to match facial part images to the first facial image;

(c) identifying the facial feature codes corresponding to the facial part images matching the first image;

(d) storing the feature codes identified in step (c) as a set of feature codes representative of the first facial image;

(e) generating a three-dimensional facial image from the set of feature codes stored in step (d), wherein the generated facial image is of a face visually similar to a face shown in the first facial image;

(f) storing a listing of sets of facial image codes, wherein each set of facial image code comprises facial feature codes corresponding to a particular individual;

(g) matching the facial feature codes identified in step (c) to the facial feature codes of at least one of said sets of facial image codes in the listing, and

(h) displaying a three-dimensional facial image generated by an assembly of the facial image parts corresponding to facial image codes of the at least one set of facial image codes matched in step (g).

11. A method as in claim 10 further comprising step (i) of comparing the facial image displayed in step (h) to the facial image generated in step (e).

12. A method as in claim 11 further comprising the step (j) of identifying an individual corresponding to the facial image generated in step (e) based on the comparison in step (i).

13. A method as in claim 10 wherein said facial part images include surface shapes of a plurality of shapes of each of a nose, eye, ear and chin.

14. A method as in claim 10 wherein the first facial image is at least one two-dimensional image.

15. A method as in claim 10 wherein the first facial image comprises a first two-dimensional facial image, and a second two-dimensional facial image orthogonal to the first two-dimensional facial image.

16. A method as in claim 10 wherein steps (f), (g) and (h) are performed on a computer remote from a computer used to perform steps (c), (d) and (e).

17. A method for generating a facial image using a memory unit, a processor and a display, wherein the method comprises the following steps:

(a) storing in the memory unit facial part images, wherein the facial part images are three-dimensional surface images of facial features;

(b) assigning a feature code to each facial part image;

(c) analyzing, with the processor, a first facial image to determine the facial part images matching the first image;

(d) identifying the feature codes for the facial part images that match the first image;

(e) storing in the memory unit the feature codes identified in step (d);

(f) generating, using the processor, a second image from a three-dimensional assembly of the facial part images stored in step (a) that correspond to the feature codes identified in step (d);

(g) displaying the second image generated in step (f);

(h) storing facial part images and their corresponding feature codes for a plurality of faces;

(i) matching the feature codes identified in step (d) to the feature codes stored in step (h), and

(j) identifying one of the faces from step (h) based on the match performed in step (i).