Back to results
US 11,343,277 B2Granted
Methods and systems for detecting spoofing of facial recognition in connection with mobile devices
AssigneeElement Inc.
InventorsFengjun Lv, Dushyant Goyal, Yang Wang, Adam Perold
Life of this patent
Estimating the term…
Abstract
Described are methods, systems, and medias for detecting spoofing of biometric identity recognition and/or validating an identity recognition match by using the camera of a mobile device, processing the user's face image or set of images at a first and second distance to generate first and second data representations, processing the first data representation into a predictive model, and comparing the data representation with the predictive model.
Claims24
A computer-implemented method for detecting spoofing of biometric identity recognition using the camera of a mobile device, the method comprising:
- a) recording, by the camera, a user's face image or set of images at a first distance;
- b) processing the user's face image or set of images at the first distance with algorithms to generate a first data representation of the user's image or set of images;
- c) processing the first data representation into a predictive model that estimates the data representations of the user's face image or set of images at other distances or orientations between the user's face and the camera, the predictive model generated by a machine learning algorithm, an image analysis algorithm, or a combination thereof, and trained using a plurality of user images and a plurality of spoofed user images;
- d) changing the distance between the user's face and the camera, wherein changing the distance comprises increasing the distance between the user's face and the camera, decreasing the distance between the user's face and the camera, changing the orientation of the camera in relation to the user's face, changing the orientation of the user's face in relation to the camera, or any combination thereof;
- e) recording, by the camera, the user's face image or set of images at a second distance;
- f) processing the user's face image or set of images at the second distance with algorithms to generate a second data representation of the user's second image or set of images;
- g) comparing the second data representation with the predictive model generated from the first data representation to determine if they match;
- h) validating an identity recognition match if the second data representation matches the predictive model generated from the first data representation;
- and
- i) comparing additional data to determine if the user presenting the face image or set of images at the first distance is a match to the user presenting the face image or set of images at the second distance if the second data representation does not match the predictive model generated from the first data representation.
7 more elements
A computer-implemented system comprising a computing device comprising at least one processor, an operating system configured to perform executable instructions, a memory, and a computer program including instructions executable by the computing device to create an application for detecting spoofing of biometric identity recognition using the camera of a mobile device, the application comprising:
- a) a software module configured to record a user's face image or set of images at a first distance;
- b) a software module configured to process the user's face image or set of images at the first distance with algorithms to generate a first data representation of the user's image or set of images;
- c) a software module configured to process the first data representation into a predictive model that estimates the data representations of the user's face image or set of images at other distances or orientations between the user's face and the camera, the predictive model generated by a machine learning algorithm, an image analysis algorithm, or a combination thereof, and trained using a plurality of user images and a plurality of spoofed user images;
- d) a software module configured to change the distance between the user's face and the camera, wherein changing the distance comprises increasing the distance between the user's face and the camera, decreasing the distance between the user's face and the camera, changing the orientation of the camera in relation to the user's face, changing the orientation of the user's face in relation to the camera, or any combination thereof;
- e) a software module configured to record the user's face image or set of images at a second distance;
- f) a software module configured to process the user's face image or set of images at the second distance with algorithms to generate a second data representation of the user's second image or set of images;
- g) a software module configured to compare the second data representation with the predictive model generated from the first data representation to determine if they match;
- h) a software module configured to validate an identity recognition match if the second data representation matches the predictive model generated from the first data representation;
- and
- i) a software module configured to compare additional data to determine if the user presenting the face image or set of images at the first distance is a match to the user presenting the face image or set of images at the second distance if the second data representation does not match the predictive model generated from the first data representation.
7 more elements
A non-transitory computer-readable storage media encoded with a computer program including instructions executable by one or more processors to create an application for detecting spoofing of biometric identity recognition using the camera of a mobile device, the application comprising:
- a) a software module configured to record a user's face image or set of images at a first distance;
- b) a software module configured to process the user's face image or set of images at the first distance with algorithms to generate a first data representation of the user's image or set of images;
- c) a software module configured to process the first data representation into a predictive model that estimates the data representations of the user's face image or set of images at other distances or orientations between the user's face and the camera, the predictive model generated by a machine learning algorithm, an image analysis algorithm, or a combination thereof, and trained using a plurality of user images and a plurality of spoofed user images;
- d) a software module configured to change the distance between the user's face and the camera, wherein changing the distance comprises increasing the distance between the user's face and the camera, decreasing the distance between the user's face and the camera, changing the orientation of the camera in relation to the user's face, changing the orientation of the user's face in relation to the camera, or any combination thereof;
- e) a software module configured to record the user's face image or set of images at a second distance;
- f) a software module configured to process the user's face image or set of images at the second distance with algorithms to generate a second data representation of the user's second image or set of images;
- g) a software module configured to compare the second data representation with the predictive model generated from the first data representation to determine if they match;
- h) a software module configured to validate an identity recognition match if the second data representation matches the predictive model generated from the first data representation;
- and
- i) a software module configured to compare additional data to determine if the user presenting the face image or set of images at the first distance is a match to the user presenting the face image or set of images at the second distance if the second data representation does not match the predictive model generated from the first data representation.
7 more elements
Dependencies are inferred from the claim text.