FACE RECOGNIZING APPARATUS AND FACE RECOGNIZING METHOD

• US 20130070973A1
• Filed: 09/12/2012
• Published: 03/21/2013
• Est. Priority Date: 09/15/2011
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Hiroshi Sukegawa, Osamu Yamaguchi
• Inventors: YAMAGUCHI, Osamu, SAITO, Hiroo, SUKEGAWA, Hiroshi
• Granted Patent: US 9,098,760 B2
• Time in Patent Office: Days
• US Class Current: 382/118
• CPC Class Codes: G06V 10/993 Evaluation of the quality of the acquired pattern
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation

Abstract

According to one embodiment, a face recognizing apparatus includes: a storage unit; an input unit; a face detector; an extractor; and a recognizing unit. The storage unit stores face feature information on a face feature of each person. The input unit receives image information including at least a face of a person. The face detector detects a face region of the face of the person from the image information received by the input unit. The extractor extracts face feature information on a face feature from the face region detected by the face detector. The recognizing unit recognizes the person in the image information received by the input unit based on the feature information extracted by the extracting unit and the face feature information stored in the storage unit.

Claims

1. A face recognizing apparatus comprising:

a storage unit that stores face feature information on a face feature of each person;

an input unit that receives image information including at least a face of a person;

a face detector that detects a face region of the face of the person from the image information received by the input unit;

an extractor that extracts face feature information on a face feature from the face region detected by the face detector; and

a recognizing unit that recognizes the person in the image information received by the input unit based on the feature information extracted by the extracting unit and the face feature information stored in the storage unit.

2. The face recognizing apparatus of claim 1, wherein

the input unit receives video data,

the face detector detects, from the input video data, a face region of a face of a same person over a plurality of frame images,

the extractor detects a face feature point for each of the frame images from the face region detected over the frame images,

the face recognizing apparatus further comprises a correcting unit configured to verify a detection accuracy of the face feature point detected for each of the frame images by comparing coordinates of the face feature point detected for one of the frame images and coordinates of the face feature point detected for other one of the frame images, and correct a detection result of the detected face feature point by performing at least one of exclusion of a face feature point failing to satisfy a predetermined detection accuracy from the detection result and correction of the coordinates of the face feature point failing to satisfy the predetermined detection accuracy, and

the recognizing unit recognizes the same person with reference to the face feature information based on the corrected detection result of the face feature point.

3. The face recognizing apparatus according to claim 2, wherein, if a difference between coordinates of the face feature point detected for one of the frame images and an average value of coordinates of face feature points detected for the frame images is greater than a predetermined value, the correcting unit determines the face feature point as failing to satisfy a predetermined detection accuracy.

4. The face recognizing apparatus according to claim 2, wherein, if a movement amount between coordinates of the face feature point detected for one of the frame images and coordinates of a face feature point detected for the nearest frame image is greater than a predetermined value, the correcting unit determines the face feature point as failing to satisfy a predetermined detection accuracy.

5. The face recognizing apparatus according to claim 2, wherein the correcting unit calculates a moving average of coordinates of the face feature points detected for the frame images, and corrects the coordinates of a face feature point failing to satisfy the predetermined detection accuracy based on the calculated moving average.

6. The face recognizing apparatus according to claim 5, wherein the correcting unit calculates a direction of a face detected based on the face feature point detected for each of the frame images, and changes a weight value used to calculate the moving average by the face feature point with the calculated face direction not oriented toward a predetermined direction and the face feature point with the calculated face direction oriented toward the predetermined direction.

7. The face recognizing apparatus according to claim 1, wherein

the input unit receives video data,

the face detector detects, from the input video data, a face region of a face of a same person over a plurality of frame images,

the face recognizing apparatus further comprises an image quality correcting unit that corrects image quality of a face region detected from a predetermined frame image based on a plurality of face images of the face regions detected over the frame images,

the extractor detects a face feature point for each of the frame images based on the face region detected over the frame images and the face region of which the image quality is corrected, and

the recognizing unit recognizes the same person with reference to the face feature information based on the detected face feature point.

8. The face recognizing apparatus according to claim 7, wherein the image quality correcting unit performs at least one of enhancing the resolution of a face region detected from the predetermined frame image, noise removal, luminance correction, brightness correction, and edge enhancement.

9. The face recognizing apparatus of claim 1, further comprising a retaining unit that retains a plurality of detecting techniques differing from each other in roughness of detection processing, each of the detecting techniques being for detecting a face feature from the face region detected by the face detector,

wherein the extracting unit extracts, by at least one of the detecting techniques retained in the retaining unit, the face feature information on the face feature from the face region detected by the face detector.

10. The face recognizing apparatus of claim 9, further comprising a selector that selects the at least one of the detecting techniques retained in the retaining unit based on a resolution of the face region detected by the face detector,

wherein the extracting unit extracts, by the at least one of the detecting techniques selected by the selector, the face feature information from the face region detected by the face detector.

11. The face recognizing apparatus of claim 9, wherein the extractor determines whether the face feature detected by one of the detecting techniques retained in the retaining unit satisfies a condition based on an average face pattern, and, if it is determined that the detected face feature does not satisfy the condition, extracts the face feature information by other one of the detecting techniques which is coarser in detection processing than the one of the detecting techniques.

12. The face recognizing apparatus of claim 1, wherein the extractor detects a position of the face feature from the face region detected by the face detector by a first detecting technique of the detecting techniques retained in the retaining unit, and extracts the face feature information from a region around the detected position by a second detecting technique of the detecting techniques which is finer in detection processing than the first detecting technique.

13. The face recognizing apparatus of claim 12, wherein the extractor detects the position of the face feature by the first detecting technique, determines whether the detected position satisfies a predetermined condition, and, if it is determined that the detected position does not satisfy the predetermined condition, performs detection processing by the second detecting technique.

14. The face recognizing apparatus of claim 9, further comprising a display that displays, upon displaying information of the person recognized by the recognizing unit, information about face recognition reliability which is based on the at least one of the detecting techniques used by the extractor.

15. A face recognizing method comprising:

receiving image information including at least a face of a person;

detecting a face region of the face of the person from the received image information;

extracting face feature information on a face feature from the detected face region; and

recognizing the person in the received image information based on the extracted feature information and face feature information indicating a feature of a face of each person.