Facial image screening method and face recognition system thereof

• US 9,971,933 B1
• Filed: 01/09/2017
• Issued: 05/15/2018
• Est. Priority Date: 01/09/2017
• Status: Active Grant

• Current Assignee: ULSee Inc.
• Original Assignee: ULSee Inc.
• Inventors: Shen, Bau-Cheng
• Primary Examiner(s): Liew, Alex
• Application Number:
  US15/401,180
• Time in Patent Office: 491 Days
• Field of Search: 382115-118
• CPC Class Codes: G06V 10/993 Evaluation of the quality of the acquired pattern
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation
• 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
• CPC Class Codes: G06V 40/173 face re-identification, e.g. recognising unknown faces across different face tracks

Abstract

A facial image screening method is provided in the present disclosure. The facial image screening method includes the following steps: tracking a plurality of feature points of at least one facial image; analyzing each feature point to acquire a region information corresponding to each feature point, wherein the region information comprises an image quality information, a face pose information, and a blocking degree information; scoring the image quality information to obtain a first arbitration score; scoring the face pose information to obtain a second arbitration score; scoring the blocking degree information to obtain a third arbitration score; generating a comprehensive quality score according to the first arbitration score, the second arbitration score, and the third arbitration; and taking the plurality of feature points of the facial image as targets to be compared with captured features of a plurality of specific persons when the comprehensive quality score exceeds a threshold value. The present disclosure has an advantage that a facial image suitable for face comparison may be selected so as to improve the recognition rate of one-to-one or one-to-many comparison.

Claims

1. A facial image screening method, comprising:

tracking a plurality of feature points of at least one facial image;

analyzing each feature point to acquire a region information corresponding to each feature point, wherein the region information comprises an image quality information, a face pose information, and a blocking degree information;

scoring the image quality information to obtain a first arbitration score;

scoring the face pose information to obtain a second arbitration score;

scoring the blocking degree information to obtain a third arbitration score;

generating a comprehensive quality score according to the first arbitration score, the second arbitration score, and the third arbitration; and

taking the plurality of feature points of the facial image as targets to be compared with captured features of a plurality of specific persons when the comprehensive quality score exceeds a threshold value.

2. The facial image screening method according to claim 1, wherein the image quality information indicates a position of the feature point tracked on the facial image.

3. The facial image screening method according to claim 1, wherein the face pose information indicates a face pose estimated in the facial image, and the face pose is defined by a range of three angles comprising a pitch angle, a roll angle, and a yaw angle.

4. The facial image screening method according to claim 1, wherein the blocking degree information indicates a blocking degree of a coverage region of the feature point tracked on the facial image.

5. The facial image screening method according to claim 1, wherein the first arbitration score is determined according to the following equation:

score_a(i)=f(image_f,image_b);

wherein image_f indicates a facial image of an unknown persons, image_b indicates an average of facial images of all specific persons in a database, f indicates an analysis of an image blur degree, and score_a(i) indicates the first arbitration score.

6. The facial image screening method according to claim 5, wherein the second arbitration score is determined according to the following equation:

score_p(i)=(P_i−

P_T)/E_p+(R_i−

P_R)/E_r+(Y_i-P_Y)/E_y;

wherein (P_i, R_i, Y_i) indicates a face pose estimated in an i-th facial image, P_T indicates the pitch angle, P_R indicates the roll angle, P_Y indicates the yaw angle, E_P indicates a tolerable error corresponding to the pitch angle, E_r indicates a tolerable error corresponding to the roll angle, E_y indicates a tolerable error corresponding to the yaw angle, and score_p(i) indicates the second arbitration score.

7. The facial image screening method according to claim 6, wherein the third arbitration score is determined according to the following equation:

8. The facial image screening method according to claim 7, wherein the comprehensive quality score is determined according to the following equation:

Comprehensive quality score=wc*score_c(i)+wp*score_p(i)+wa*score_a(i);

wherein we indicates a third weighting value corresponding to the third arbitration score score_c(i), wp indicates a second weighting value corresponding to the second arbitration score score_p(i), and wa indicates a first weighting value corresponding to the first arbitration score score_a(i).

9. A face recognition system, the face recognition system comprising:

a database, used for storing captured features of a plurality of specific persons;

a face detection and analysis module, used for tracking a plurality of feature points of at least one facial image of an unknown person, and for analyzing each feature point to acquire a region information corresponding to each feature point, wherein the region information comprises an image quality information, a face pose information, and a blocking degree information;

a score module, used for scoring the image quality information to obtain a first arbitration score, for scoring the face pose information to obtain a second arbitration score, and for scoring the blocking degree information to obtain a third arbitration score;

a quality estimation module, used for generating a comprehensive quality score according to the first arbitration score, the second arbitration score, and the third arbitration; and

a comparison module, used for taking the plurality of feature points of the facial image as targets to be compared with captured features of a plurality of specific persons when the comprehensive quality score exceeds a threshold value.

10. The face recognition system according to claim 9, wherein the image quality information indicates a position of the feature point tracked on the facial image.

11. The face recognition system according to claim 9, wherein the face pose information indicates a face pose estimated in the facial image, and the face pose is defined by a range of three angles comprising a pitch angle, a roll angle, and a yaw angle.

12. The face recognition system according to claim 9, wherein the blocking degree information indicates a blocking degree of a coverage region of the feature point tracked on the facial image.

13. The face recognition system according to claim 9, wherein the first arbitration score is determined according to the following equation:

score_a(i)=f(image_f,image_b);

wherein image_f indicates a facial image of an unknown persons, image_b indicates an average of facial images of all specific persons in a database, f indicates an analysis of an image blur degree, and score_a(i) indicates the first arbitration score.

14. The face recognition system according to claim 13, wherein the second arbitration score is determined according to the following equation:

score_p(i)=(P_i−

P_T)/E_p+(R_i−

P_R)/E_r+(Y_i−

P_Y)/E_y;

wherein (P_i, R_i, Y_i) indicates a face pose estimated in an i-th facial image, P_T indicates the pitch angle, P_R indicates the roll angle, P_Y indicates the yaw angle, E_P indicates a tolerable error corresponding to the pitch angle, E_r indicates a tolerable error corresponding to the roll angle, E_y indicates a tolerable error corresponding to the yaw angle, and score_p(i) indicates the second arbitration score.

15. The face recognition system according to claim 14, wherein the third arbitration score is determined according to the following equation:

16. The face recognition system according to claim 15, wherein the comprehensive quality score is determined according to the following equation:

Comprehensive quality score=wc*score_c(i)+wp*score_p(i)+wa*score_a(i);

wherein we indicates a third weighting value corresponding to the third arbitration scorescore_c(i), wp indicates a second weighting value corresponding to the second arbitration score score_p(i), and wa indicates a first weighting value corresponding to the first arbitration score score_a(i).