Detection of identity changes during facial recognition enrollment process
A device with a camera may utilize an enrollment process to capture images of an authorized user to enroll the user for a facial recognition authorization process. The enrollment process may include one or more processes that identify if an identity of the authorized user (e.g., the subject of the enrollment process) has switched during the enrollment process. The processes may include detection and verification of the switch in identities by comparing features of subjects in images as the images are captured during the enrollment process. If the identity of the subject is determined to be switched from the authorized user during the enrollment process, the enrollment process may be restarted. Additionally, clustering of feature vectors from the enrollment images may be used to remove outlying feature vectors that may be generated from one or more images of a subject other than the authorized user.
1. A method, comprising:
capturing a first image using a camera located on a device, the device comprising a computer processor and a memory, wherein the first image comprises an image captured while illuminating a subject in the image with infrared illumination from a projector located on the device, the first image being captured during a facial recognition enrollment authorization process;
determining whether the subject in the first image is a different subject than a reference subject from one or more images captured during the facial recognition enrollment authorization process prior to the first image being captured;
in response to the subject in the first image being determined to be different than the reference subject, capturing one or more second images using the camera located on the device;
determining whether a subject in the second images is the same subject as the subject in the first image; and
restarting the facial recognition enrollment authorization process in response to the subject in the second images being determined to be the same subject as the subject in the first image.
2. The method of claim 1 , wherein the infrared illumination comprises flood infrared illumination and speckle pattern infrared illumination.
3. The method of claim 1 , wherein determining whether the subject in the first image is a different subject than the reference subject from the one or more images captured during the facial recognition enrollment authorization process prior to the first image being captured comprises:
encoding the first image to generate first feature vectors defining features of the subject in the first image in a feature space; and
comparing the first feature vectors in the feature space to reference feature vectors for the reference subject.
4. The method of claim 1 , wherein determining whether the subject in the second images is the same subject as the subject in the first image comprises:
encoding the first image to generate first feature vectors defining features of the subject in the first image in a feature space;
encoding the second images to generate second feature vectors defining features of the subject in the second images in the feature space; and
comparing the first feature vectors to the second feature vectors in the feature space.
5. The method of claim 1 , further comprising capturing a predetermined number of second images before determining whether the subject in the second images is the same subject as the subject in the first image.
6. The method of claim 5 , further comprising determining whether the subject in the predetermined number of second images comprises a different subject than the reference subject before restarting the facial recognition enrollment authorization process.
7. The method of claim 1 , further comprising maintaining the reference subject in the facial recognition enrollment authorization process when the subject in the second images is a different subject from the subject in the first image.
8. A method, comprising:
capturing a first set of images using a camera located on a device, the device comprising a computer processor and a memory, wherein the first set of images comprises one or more images captured while illuminating a face of a first subject in the images with infrared illumination from a projector located on the device, the first set of images being captured during a facial recognition enrollment authorization process;
encoding the first set of images to generate first feature vectors defining features of the first subject in the first set of images in a feature space;
capturing a second set of images using the camera located on the device, wherein the second set of images comprises one or more images captured while illuminating a face of a second subject in the images with the infrared illumination from the projector located on the device, the second set of images being captured during the facial recognition enrollment authorization process;
encoding images in the second set of images as the images in the second set of images are captured, wherein encoding the images in the second set of images generates second feature vectors defining features of the second subject in the feature space;
comparing at least some of the second feature vectors to one or more of the first feature vectors; and
determining during the facial recognition enrollment authorization process, whether the second subject is a different subject from the first subject based on the comparison of at least some of the second feature vectors to one or more of the first feature vectors.
9. The method of claim 8 , wherein comparing at least some of the second feature vectors to one or more of the first feature vectors comprises:
assessing an average distance between the first feature vectors and the second feature vectors; and
comparing the average distance to at least one threshold to determine whether the second subject is a different subject from the first subject.
10. The method of claim 8 , wherein comparing at least some of the second feature vectors to one or more of the first feature vectors comprises:
determining a matching score between the at least some second feature vectors and the one or more first feature vectors;
assessing whether the matching score is below a threshold; and
in response to the matching score being below the threshold, comparing the second feature vectors of at least one second image to the second feature vectors of at least one other second image captured prior to the at least one second image.
11. The method of claim 8 , further comprising assessing clustering of the first feature vectors and the second feature vectors in the feature space, and removing feature vectors from the feature space that are outliers from the assessed clusters in the feature space, the feature space then being used as a feature space for facial recognition enrollment authorization using the device.
12. The method of claim 8 , further comprising restarting the facial recognition enrollment authorization process in response to the second subject being determined to be a different subject from the first subject.
13. The method of claim 12 , further comprising providing a notification on a display of the device that the facial recognition enrollment authorization process is being restarted.
14. The method of claim 8 , wherein the second set of images comprises a set of images captured immediately following the first set of images.
15. The method of claim 8 , wherein the infrared illumination comprises flood infrared illumination and speckle pattern infrared illumination.
16. A device, comprising:
a camera;
at least one illuminator providing infrared illumination;
circuitry coupled to the camera and the illuminator, wherein the circuitry is configured to:
capture a first image using the camera, wherein the first image comprises an image captured while illuminating a subject in the image with infrared illumination from the at least one illuminator, the first image being captured during a facial recognition enrollment authorization process;
determine whether the subject in the first image is a different subject than a reference subject from one or more images captured during the facial recognition enrollment authorization process prior to the first image being captured;
in response to the subject in the first image being determined to be different than the reference subject, capture one or more second images using the camera;
determine whether a subject in the second images is the same subject as the subject in the first image; and
restart the facial recognition enrollment authorization process in response to the subject in the second images being determined to be the same subject as the subject in the first image.
17. The device of claim 16 , wherein the circuitry is configured to provide a notification on a display of the device that the facial recognition enrollment authorization process is being restarted.
18. The device of claim 16 , wherein the at least one illuminator comprises a flood infrared illuminator and a speckle pattern infrared illuminator.
19. The device of claim 16 , wherein the circuitry is configured to authenticate the reference subject to use the device in response to the subject in the second images being a different subject as the subject in the first image.
20. The device of claim 16 , wherein the circuitry is configured to authenticate the reference subject to use the device in response to the subject in the first image being the same subject as the reference subject from the one or more images captured prior to the first image being captured.