Facial recognition to positively identify a live person subject

1. A method to authenticate a live person subject, comprising:

receiving an authentication request from a user;

generating, in response to the authentication request, a sequence of instructions instructing the user to point a face of the user toward a sequence of facial directions, wherein the sequence of facial directions are randomly generated using a random sequence generation algorithm, wherein the sequence of facial directions correspond to a sequence of reference facial images of the live person subject;

presenting the sequence of instructions to the user, comprising:

presenting a first instruction at a first time point, wherein the first instruction instructs the user to point the face toward a first facial direction; and

presenting a second instruction at a second time point, wherein the second instruction instructs the user to point the face toward a second facial direction,

wherein the first time point and the second time point are determined based on instruction delays randomly generated using the random sequence generation algorithm;

capturing, while presenting the sequence of instructions to the user, a sequence of live-captured facial images (LCFIs) based on a pre-determined frame rate; and

generating an authentication result identifying the user as the live person subject by at least:

matching a LCFI in the sequence of LCFIs to a first reference facial image of the live person subject toward the first facial direction; and

matching another LCFI in the sequence of LCFIs to a second reference facial image of the live person subject toward the second facial direction; and

validating each LCFI in the sequence of LCFIs based on a pre-determined criterion.

2. The method of claim 1 , further comprising:

obtaining, during a learning phase prior to receiving the authentication request, a plurality of reference facial images of the live person subject corresponding to a plurality of facial directions of the live person subject,

wherein the first reference facial image and the second reference facial image are obtained based on the plurality of reference facial images.

3. The method of claim 2 , further comprising:

selecting the first reference facial image and the second reference facial image from the plurality of reference facial images based on the first facial direction and the second facial direction,

wherein plurality of facial directions comprise the first facial direction and the second facial direction.

4. The method of claim 1 , further comprising:

selecting the LCFI from the sequence of LCFIs based on the first time point when the first instruction is presented during presenting the sequence of instruction to the user; and

selecting the another LCFI from the sequence of LCFIs based on the second time point when the second instruction is presented during presenting the sequence of instruction to the user.

5. The method of claim 1 , wherein validating each LCFI in the sequence of LCFIs comprises:

generating a difference measure between two consecutive LCFIs in each consecutive LCFI pair in the sequence of LCFIs; and

determining that each difference measure of all consecutive LCFI pairs in the sequence of LCFIs is less than a pre-determined threshold.

6. The method of claim 1 ,

wherein the pre-determined frame rate is a constant frame rate.

7. The method of claim 1 ,

wherein the pre-determined frame rate is a variable frame rate.

8. A system for authenticating a live person subject, comprising:

a computer processor;

a random instruction module executing on the computer processor and configured to:

receive an authentication request from a user;

generate, in response to the authentication request, a sequence of instructions instructing the user to point a face of the user toward a sequence of facial directions, wherein the sequence of facial directions are randomly generated using a random sequence generation algorithm, wherein the sequence of facial directions correspond to a sequence of reference facial images of the live person subject; and

present the sequence of instructions to the user, comprising:

presenting a first instruction at a first time point, wherein the first instruction instructs the user to point the face toward a first facial direction; and

presenting a second instruction at a second time point, wherein the second instruction instructs the user to point the face toward a second facial direction,

wherein the first time point and the second time point are determined based on instruction delays randomly generated using the random sequence generation algorithm;

an image sensor coupled to the computer processor and configured to capture, while the sequence of instructions are being presented to the user, a sequence of live-captured facial images (LCFIs) based on a pre-determined frame rate;

an LCFI analyzer executing on the computer processor and configured to generate an authentication result identifying the user as the live person subject by at least:

matching a LCFI in the sequence of LCFIs to a first reference facial image of the live person subject toward the first facial direction; and

matching another LCFI in the sequence of LCFIs to a second reference facial image of the live person subject toward the second facial direction; and

validating each LCFI in the sequence of LCFIs based on a pre-determined criterion; and

a repository configured to store the sequence of LCFIs, the first reference facial image, and the second reference facial image.

9. The system of claim 8 , further comprising a reference facial image collector configured to:

obtain, during a learning phase prior to receiving the authentication request, a plurality of reference facial images of the live person subject corresponding to a plurality of facial directions of the live person subject,

wherein the first reference facial image and the second reference facial image are obtained based on the plurality of reference facial images, and

wherein the repository is further configured to store the plurality of reference facial images.

10. The system of claim 9 , wherein the LCFI analyzer is further configured to:

select the first reference facial image and the second reference facial image from the plurality of reference facial images based on the first facial direction and the second facial direction,

wherein plurality of facial directions comprise the first facial direction and the second facial direction.

11. The system of claim 8 , wherein the LCFI analyzer is further configured to:

select the LCFI from the sequence of LCFIs based on the first time point when the first instruction is presented during presenting the sequence of instruction to the user; and

select the another LCFI from the sequence of LCFIs based on the second time point when the second instruction is presented during presenting the sequence of instruction to the user,

wherein the random instruction module is further configured to generate the instruction delays for presenting the sequence of instructions to the user.

12. The system of claim 8 , wherein validating each LCFI in the sequence of LCFIs comprises:

generating a difference measure between two consecutive LCFIs in each consecutive LCFI pair in the sequence of LCFIs; and

determining that each difference measure of all consecutive LCFI pairs in the sequence of LCFIs is less than a pre-determined threshold.

13. The system of claim 8 ,

wherein the pre-determined frame rate is a constant frame rate.

14. The system of claim 8 , wherein the random instruction module is further configured to:

generate, using the random sequence generation algorithm, the pre-determined frame rate as a variable frame rate.

15. A non-transitory computer readable medium embodying instructions for authenticating a live person subject, the instructions when executed by a processor comprising functionality for:

receiving an authentication request from a user;

generating, in response to the authentication request, a sequence of instructions instructing the user to point a face of the user toward a sequence of facial directions, wherein the sequence of facial directions are randomly generated using a random sequence generation algorithm, wherein the sequence of facial directions correspond to a sequence of reference facial images of the live person subject;

presenting the sequence of instructions to the user, comprising:

presenting a first instruction at a first time point, wherein the first instruction instructs the user to point the face toward a first facial direction; and

presenting a second instruction at a second time point, wherein the second instruction instructs the user to point the face toward a second facial direction,

wherein the first time point and the second time point are determined based on instruction delays randomly generated using the random sequence generation algorithm;

capturing, while presenting the sequence of instructions to the user, a sequence of live-captured facial images (LCFIs) based on a pre-determined frame rate; and

generating an authentication result identifying the user as the live person subject by at least:

matching a LCFI in the sequence of LCFIs to a first reference facial image of the live person subject toward the first facial direction; and

matching another LCFI in the sequence of LCFIs to a second reference facial image of the live person subject toward the second facial direction;

and

validating each LCFI in the sequence of LCFIs based on a pre-determined criterion.

16. The non-transitory computer readable medium of claim 15 , the instructions when executed by the processor further comprising functionality for:

obtaining, during a learning phase prior to receiving the authentication request, a plurality of reference facial images of the live person subject corresponding to a plurality of facial directions of the live person subject,

wherein the first reference facial image and the second reference facial image are obtained based on the plurality of reference facial images.

17. The non-transitory computer readable medium of claim 15 , the instructions when executed by the processor further comprising functionality for:

selecting the first reference facial image and the second reference facial image from the plurality of reference facial images based on the first facial direction and the second facial direction,

wherein plurality of facial directions comprise the first facial direction and the second facial direction.

18. The non-transitory computer readable medium of claim 15 , the instructions when executed by the processor further comprising functionality for:

selecting the LCFI from the sequence of LCFIs based on the first time point when the first instruction is presented during presenting the sequence of instruction to the user; and

selecting the another LCFI from the sequence of LCFIs based on the second time point when the second instruction is presented during presenting the sequence of instruction to the user.

19. The non-transitory computer readable medium of claim 15 , wherein validating each LCFI in the sequence of LCFIs comprises:

generating a difference measure between two consecutive LCFIs in each consecutive LCFI pair in the sequence of LCFIs; and

determining that each difference measure of all consecutive LCFI pairs in the sequence of LCFIs is less than a pre-determined threshold.