Process for updating templates used in facial recognition

Abstract

Templates used for a facial recognition process for authentication of a user to use a device may be updated by the device as features of the user change over time. Features of the user may gradually change over time due to changes such as facial hair changes, haircuts, gaining/losing weight, and/or aging. Updating the templates used for the facial recognition process may allow the user to continue being authenticated as features of the user change without the need for additional enrollments of the user.

1 - 20 . (canceled) 
     
     
         21 . A method, comprising:
 obtaining an image of a face of a user using a camera located on a device in response to an authentication attempt initiated by the user, the device comprising a computer processor and a memory;   encoding the image to generate a first feature vector representing one or more features of the user in the image;   authorizing the user in the image as an authenticated user of the device based on a matching score between the first feature vector and one or more reference templates stored in the memory being above a first threshold, wherein the one or more reference templates include at least one static template; and   storing the first feature vector as a dynamic template in the memory in response to the matching score being above a second threshold, wherein:
 when a number of dynamic templates stored in the memory is below a predetermined number of dynamic templates allowed in the memory, the first feature vector is stored as a new dynamic template in the memory; and 
 when the number of dynamic templates stored in the memory is at the predetermined number of dynamic templates allowed in the memory, the first feature vector replaces an existing dynamic template in the memory when the first feature vector is a closer match to the one or more reference templates and existing dynamic templates in the memory as compared to the existing dynamic template being replaced. 
   
     
     
         22 . The method of  claim 21 , wherein the second threshold is higher than the first threshold. 
     
     
         23 . The method of  claim 21 , wherein the first feature vector is a closer match when the first feature vector is less of an outlier than the existing temporary dynamic template being replaced. 
     
     
         24 . The method of  claim 23 , further comprising determining whether the first feature vector is less of the outlier based on statistical analysis of the first feature vector, the one or more reference templates, and the existing dynamic templates. 
     
     
         25 . The method of  claim 21 , further comprising:
 obtaining one or more reference images of the face of the user using the camera;   encoding facial features of the user from the one or more reference images to generate one or more reference feature vectors; and   storing the one or more reference feature vectors as the one or more reference templates in the memory of the device.   
     
     
         26 . The method of  claim 21 , wherein the at least one static template includes a template that is not deleted or changed in the memory of the device until the device is reset or a user determines to replace the one or more reference templates in the memory of the device. 
     
     
         27 . The method of  claim 21 , further comprising:
 obtaining one or more additional images using the camera in response to additional authentication attempts initiated on the device;   storing the first feature vector as the dynamic template in the memory in response to the matching score being above the second threshold and the first feature vector meeting at least one criteria for the additional images in comparison to the at least one static template.   
     
     
         28 . The method of  claim 27 , further comprising:
 encoding the additional images to generate second feature vectors representing one or more features of the user in the additional image;   generating a first set of matching scores between the second feature vectors and the first feature vector; and   generating a second set of matching scores between the second features vectors and the at least one static template;   wherein the at least one criteria includes a number of matching scores above the first threshold in the first set of matching scores being greater than a number of matching scores above the first threshold in the second set of matching scores.   
     
     
         29 . A device, comprising:
 a camera;   a processor coupled to the camera;   a memory coupled to the processor; and   circuitry coupled to the camera, wherein the circuitry is configured to:
 obtain an image of a face of a user using the camera in response to an authentication attempt initiated by the user; 
 encode the image to generate a first feature vector representing one or more features of the user in the image; 
 authorize the user in the image as an authenticated user of the device based on a matching score between the first feature vector and one or more reference templates stored in the memory being above a first threshold, wherein the one or more reference templates include at least one static template; and 
 store the first feature vector as a dynamic template in the memory in response to the matching score being above a second threshold, wherein:
 when a number of dynamic templates stored in the memory is below a predetermined number of dynamic templates allowed in the memory, the first feature vector is stored as a new dynamic template in the memory; and 
 when the number of dynamic templates stored in the memory is at the predetermined number of dynamic templates allowed in the memory, the first feature vector replaces an existing dynamic template in the memory when the first feature vector is a closer match to the one or more reference templates and existing dynamic templates in the memory as compared to the existing dynamic template being replaced. 
 
   
     
     
         30 . The device of  claim 29 , wherein the device further comprises at least one illuminator configured to provide infrared illumination directed towards the user when obtaining the image of the face of the user using the camera. 
     
     
         31 . The device of  claim 29 , wherein the camera includes an infrared sensor. 
     
     
         32 . The device of  claim 29 , wherein the second threshold is higher than the first threshold. 
     
     
         33 . The device of  claim 29 , wherein the circuitry is configured to enroll the authenticated user on the device in response to the authenticated user providing at least one additional authentication protocol on the device, and wherein the circuitry is configured to enroll the authenticated user by:
 obtaining one or more reference images of the face of the user using the camera;   encoding facial features of the user from the one or more reference images to generate one or more reference feature vectors; and   storing the one or more reference feature vectors as the one or more reference templates in the memory of the device.   
     
     
         34 . The device of  claim 33 , wherein the at least one additional authentication protocol comprises entering a passcode for the authenticated user. 
     
     
         35 . The device of  claim 29 , wherein the circuitry is configured to update the first threshold when the first feature vector is stored as the dynamic template in the memory. 
     
     
         36 . The device of  claim 29 , wherein the circuitry is configured to:
 obtain one or more additional images using the camera in response to additional authentication attempts initiated on the device;   store the first feature vector as the dynamic template in the memory in response to the matching score being above the second threshold and the first feature vector meeting at least one criteria for the additional images in comparison to the at least one static template.   
     
     
         37 . A non-transitory computer-readable medium having instructions stored thereon that are executable by a computing device to perform operations, comprising:
 obtaining an image of a face of a user using a camera located on the computing device in response to an authentication attempt initiated by the user;   encoding the image to generate a first feature vector representing one or more features of the user in the image;   authorizing the user in the image as an authenticated user of the device based on a matching score between the first feature vector and one or more reference templates stored in a memory of the computing device being above a first threshold, wherein the one or more reference templates include at least one static template; and   storing the first feature vector as a dynamic template in the memory in response to the matching score being above a second threshold, wherein:
 when a number of dynamic templates stored in the memory is below a predetermined number of dynamic templates allowed in the memory, the first feature vector is stored as a new dynamic template in the memory; and 
 when the number of dynamic templates stored in the memory is at the predetermined number of dynamic templates allowed in the memory, the first feature vector replaces an existing dynamic template in the memory when the first feature vector is a closer match to the one or more reference templates and existing dynamic templates in the memory as compared to the existing dynamic template being replaced. 
   
     
     
         38 . The non-transitory computer-readable medium of  claim 37 , wherein the image obtained includes an image of the user captured while illuminating the user with flood infrared illumination. 
     
     
         39 . The non-transitory computer-readable medium of  claim 37 , wherein the image obtained includes an image of the user captured while illuminating the user with patterned infrared illumination. 
     
     
         40 . The non-transitory computer-readable medium of  claim 37 , wherein authorizing the user in the image as the authenticated user of the device allows the authenticated user to have access to a selected functionality of the device.