Face Forensics (f2) can be used in both one-to-one (verification) and one-to-many (identification) modes. It is available both as an application and as a .NET SDK.
These applications all relate to confirming that the person presenting a credential such as a passport, driver’s licence, social security card, bank card, access card, ID card, etc, is the person to whom it was issued.
The process is fairly straightforward – when the card is issued to the individual their facial image is captured in a controlled environment, i.e. they’re looking at the camera, they’re not smiling, the lighting and background are standard, etc. The image can be stored on a server and optionally on the card itself (on a chip or 2-D barcode). f2 encodes numerous characteristics of the image and the resulting string can also be stored.
When the card is subsequently presented the facial image is captured again and encoded. The resulting string is compared with the original – if they match within a pre-defined tolerance the presenter is deemed to be the legitimate holder of the card. For really secure authentication this can be supplemented in a layered approach with a second biometric such as fingerprints.
These applications involve searching all records in a database to determine if a record already exists for a given facial image. It returns a list of likely matches in order of matching probability, so that an officer can make the final decision on whether any image is of the individual or not. This process takes seconds, so is practical in a real-time environment such as border control. The purpose of one-to-many matching is not to positively identify an individual (for example it will not, by definition, distinguish between identical twins) but rather to narrow down the possible matches to a number that an officer can easily handle.
Specific applications include the following:
Identification of individuals in a database or on a watchlist
f2 can take a photo of any individual, encode it and match it against a database which includes facial images. This can be done in real-time
Face recognition is the primary biometric mandated by the International Civil Aviation Organization (ICAO) for ePassports. f2 is ideal for this as unlike alternative systems it can be readily configured to work with an existing passport database. This can be done without the assistance or even knowledge of the database supplier. Third-party databases are accessed in read-only mode so data integrity is assured. While this is a 1:1 matching process, to check if the individual has another record under a different name requires 1:N. f2 is also available in a 64-bit version, enabling it to work with databases containing many millions of records.
Drivers Licenses & Entitlement Document Databases
f2 can readily detect instances where an individual has applied for multiple entitlement documents under different names. The program is simply pointed at the images in the relevant database and each image is compared against every other image. This is a time-consuming process so the program runs as a background task, enabling other programs to be run on the same computer at the same time. Where a possible match is identified the two facial images are displayed side-by-side for an officer to check.
Real-time Face Screening at Checkpoints
The ability of f2 to detect multiple faces in a video stream and then extract them means that it can be used to watch people coming through a checkpoint and match their faces against a watchlist. Where the match is above a predefined threshold, f2 will display both faces on the officer’s screen so that a visual comparison can be made. In order that there may not be too many false positives, i.e. where the two faces are similar but not of the same person, the threshold is generally set relatively high.
Missing Children – Webpages & hard-drives
For law enforcement f2 has the ability to be pointed at selected websites that are known to contain images of missing & abused children to detect whether any of the faces match those on a watchlist. The program runs as a background application, which means that once started it can be left running while other applications are being run on the computer. When a computer is suspected of holding illegal images f2 can search every image on the computer, isolate any faces, and display matches against a watchlist.