Glossary
A FaceScan is what Verisoul captures on a client device when verifying a user identity. A FaceScan includes two important data points to help ensure liveness and uniqueness.
- Extracted 3D vectors mapped to represent a user's facial features and landmarks
- It is crucial the vectors remain three dimensional to prevent common spoofs like presenting a two dimensional picture of a user
- Note: this does contain some image data but it is immediately encrypted and used only for verification; it is never stored
- Liveness data that captures how a user's vectorized features changed over the duration of a FaceScan
- Distortions of a face as it moves are measured and help to ensure the user is a real, living person
A FaceScan is used as the basis to create a non-reversible FacePrint and are never stored.
A FacePrint is fundamentally an array of numbers. It's a n-dimensional feature vector that describes numerical features of some object in pattern recognition in machine learning. A FacePrint contains no human visible face data and non-reversible meaning it cannot be used to reconstruct a human viewable image of a face.
A VoicePrint is fundamentally a buffer of data. It's an opaque sequence of bytes that corresponds to the speech of the enrolled person to be used in verification process. It encodes unique attributes about a user's speech like tonality, clarity, etc. A VoicePrint is anonymized such that it contains no voice recording data and is non-reversible.
Authentication whereby an identity is captured and compared with all previously captured identities. The search results are used to establish the identity of a person. In short, a 1:N Identification answers the question:
- Who is this person and are they unique?
Examples of a 1:N Identification include any place that requires Social Security Number (SSN), ID/Passport issuance, Verisoul's Liveness & Uniqueness solution, etc.
Authentication whereby an identity is captured and compared with a single previously captured identity. In short, a 1:1 Verification answers the question:
- Is this person who they claim to be?
Examples of a 1:1 Verification are Apple's FaceID, CLEAR travel biometrics, inputting a PIN at a bank ATM, etc. This approach is usually made more secure by adding an additional "factor" of authentication like having possession of a phone, password, etc.
Last modified 1yr ago