FACE RECOGNITION METHOD

• US 20170236002A1
• Filed: 02/16/2016
• Published: 08/17/2017
• Est. Priority Date: 02/15/2016
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: OH, Sang Yoon
• Granted Patent: US 9,898,648 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06V 40/171 Local features and components; Facial parts eye characteristics G06V40/18; Occluding parts, e.g. glasses; Geometrical relationships
• CPC Class Codes: G06V 40/172 Classification, e.g. identification

Abstract

A face recognition method of the present disclosure includes configuring aggregations of feature data that include a plurality of feature data of faces and match to a plurality of personnel data; extracting from an input image a plurality of input feature data that correspond to the feature data and that is equal to or more than a critical value; comparing an aggregation of input feature data that includes the input feature data with each of the pre-stored aggregations of feature data, and selecting from the aggregations of feature data an aggregation of feature data having the greatest similarity with the aggregation of the input feature data; and identifying a person on the image based on personnel data that matches the aggregation of feature data having the greatest similarity.

Claims

1. A face recognition method comprising:

configuring aggregations of feature data that include a plurality of feature data of faces and match to a plurality of personnel data;

extracting from an input image a plurality of input feature data that correspond to the feature data and that is equal to or more than a critical value;

comparing an aggregation of input feature data that includes the input feature data with each of the pre-stored aggregations of feature data, and selecting from the aggregations of feature data an aggregation of feature data having the greatest similarity with the aggregation of the input feature data; and

identifying a person on the image based on personnel data that matches the aggregation of feature data having the greatest similarity.

2. The method of claim 1,

wherein the extracting the plurality of input feature data involves setting a value of the feature data that is unrecognizable from the input image or that is less than the critical value to zero (0).

3. The method of claim 2,

wherein the selecting an aggregation of feature data having the greatest similarity comprises:

computing values of comparison between the input feature data and the feature data for each of the aggregations of feature data;

computing a sum of the computed values of comparison; and

selecting an aggregation with the smallest sum of values of comparison as the aggregation of feature data having the greatest similarity.

4. The method of claim 3,

wherein the computing the values of comparison involves setting the values of comparison between the feature data that is unrecognizable and the feature data corresponding to the unrecognizable feature data to zero (0).

5. The method of claim 3,

further comprising determining whether or not the sum of the values of comparison is below a reference value.

6. The method of claim 5,

wherein the identifying a person on the image involves, when the sum of the values of comparison of the aggregation of feature data having the greatest similarity is equal to or above the reference value, determining that there is no personnel data corresponding to the person on the image.

7. The method of claim 1,

wherein the plurality of aggregations of feature data of a same person are pre-stored in a feature data DB.

8. The method of claim 1,

wherein the feature data is a distance between feature points or a ratio of widths of certain areas.

9. The method of claim 8,

wherein the distance between feature points or the widths of certain areas is calculated in pixel units.

10. The method of claim 8,

wherein the extracting input feature data involves recognizing a face outline or a feature area of the face from the image.

11. The method of claim 10,

wherein the feature data comprises:

first feature data that is a ratio of a width of a right eye and a first face width;

second feature data that is a ratio of a width of a nose and a second face width;

third feature data that is a ratio of a width of a left eye and the width of the nose;

fourth feature that is a ratio of a width of a mouth and a third face width;

fifth feature that is a ratio of the width of the right eye and the width of the mouth; and

sixth feature data that is a ratio of the width of the nose and the width of the mouth.

12. The method of claim 11,

wherein the width of the right eye is defined as a maximum width of a right eye area in a width direction.

13. The method of claim 11,

wherein the width of the left eye is defined as a maximum width of a left eye area in a width direction.

14. The method of claim 12,

wherein the first face width is defined as a distance between the points at which a horizontal line from a central point of the right eye area and a central point of the left eye area intersects with the face outline.

15. The method of claim 11,

wherein the width of the nose is defined as a maximum width of a nose area in a width direction.

16. The method of claim 15,

wherein the second face width is defined as a distance between the points at which a horizontal line extended from the maximum width of the nose area in the width direction intersects with the face outline.

17. The method of claim 11,

wherein the width of the mouth is defined as a maximum width of a mouth area in a width direction.

18. The method of claim 17,

wherein the third face width is defined as a distance between the points at which a horizontal line extended from the maximum width of the mouth area in the width direction intersects with the face outline.