SYSTEM AND METHOD FOR VALIDATION OF FACE DETECTION IN ELECTRONIC IMAGES

• US 20100040288A1
• Filed: 08/18/2008
• Published: 02/18/2010
• Est. Priority Date: 08/18/2008
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Tec Corporation (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Tec Corporation (Toshiba Corporation)
• Inventors: YEN, Jonathan, Quach, Tony
• Granted Patent: US 8,238,604 B2
• Time in Patent Office: Days
• US Class Current: 382/195
• CPC Class Codes: G06F 18/217 Validation; Performance evaluation; Active pattern learning techniques
• CPC Class Codes: G06V 10/776 Validation; Performance evaluation
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

The subject application is directed to a system and method for validation of face detection in electronic images. Image data is first received along with at least one image portion that includes a possible facial depiction. Eye position data, nose position data, and mouth position data are also received. A reference point at a central location of the at least one image portion is then isolated. A width of the image portion is then isolated, and a facial region is isolated in accordance with the eye, nose, and mouth position data. The eye distance is then determined from the received eye position data. The isolated facial region data is then tested against the reference point and eye distance is tested against a width of the image portion. An output is generated corresponding to the accuracy of isolated facial region in accordance with the tests.

Claims

1. A facial region detection system, comprising:

an image data input;

a detected facial area candidate data input;

a facial area center point detector operable on a facial area candidate;

a nose area data input;

an eye area data input;

a mouth area data input;

a facial region calculator operable on received nose area data, received eye area data and received mouth area data;

a coincidence comparator operable on a data corresponding to a facial area center point and a facial region;

a measurement comparator operable on eye distance data corresponding to a distance between two eyes as represented by eye data and on facial area distance data corresponding to a side length associated with the facial area candidate;

a facial area detection signal generator operable in accordance with an output of the coincidence comparator and an output of the measurement comparator.

2. The facial detection system of claim 1, wherein the coincidence comparator generates an output in accordance with an overlap of the facial area center point and the facial region.

3. The facial detection system of claim 2, wherein the facial area candidate is rectangular.

4. The facial detection system of claim 3, wherein the facial region is a convex shape.

5. A method for validation of face detection in electronic images, comprising the steps of:

receiving image data;

receiving at least one image portion of received image data, including at least a portion of a possible facial depiction;

receiving eye position data corresponding to at least one eye position in the at least one image portion;

receiving nose position data corresponding to a nose position in the at least one image portion;

receiving mouth position data corresponding to a mouth position in the at least one image portion;

isolating a reference point at a central location of the at least one image portion;

isolating a width of the at least one image portion;

isolating a facial region in accordance with eye position data, nose position data, and mouth position data;

determining eye distance from received eye position data;

testing isolated facial region data against the reference point;

testing eye distance against a width of the image portion;

generating an output corresponding to the accuracy of isolated facial region in accordance with an output of the testing steps.

6. The method of claim 5, wherein the at least one image portion is defined as a generally rectangular sub-region of the image data.

7. The method of claim 6, wherein the step of isolating the facial region includes the step of defining the facial region as a convex sub-region of the image data.

8. The method of claim 7, further comprising the step of isolating a rotational factor in accordance with a result of the testing step.

12. The system of claim 7, further comprising means adapted for isolating a rotational factor in accordance with a result of the testing means.

9. A system for validation of face detection in electronic images, comprising the steps of:

means adapted for receiving image data;

means adapted for receiving at least one image portion of received image data, including at least a portion of a possible facial depiction;

means adapted for receiving eye position data corresponding to at least one eye position in the at least one image portion;

means adapted for receiving nose position data corresponding to a nose position in the at least one image portion;

means adapted for receiving mouth position data corresponding to a mouth position in the at least one image portion;

means adapted for isolating a reference point at a central location of the at least one image portion;

means adapted for isolating a width of the at least one image portion;

means adapted for isolating a facial region in accordance with eye position data, nose position data, and mouth position data;

means adapted for determining eye distance from received eye position data;

means adapted for testing isolated facial region data against the reference point;

means adapted for testing eye distance against a width of the image portion;

means adapted for generating an output corresponding to the accuracy of isolated facial region in accordance with an output of the testing means.

10. The system of claim 9, wherein the at least one image portion is defined as a generally rectangular sub-region of the image data.

11. The system of claim 10, wherein the means adapted for isolating the facial region includes means adapted for defining the facial region as a convex sub-region of the image data.