Facial Recognition

Facial Recognition

Dectel is working with Identix Inc to provide working solutions using automatic facial recognition technology. With this technology, Dectel is able to provide solutions in many different areas of business. The technology can be used to perform simple verification procedures to identification across large databases.

  • Face Recognition Software
  • Multi-Camera Solutions
  • Retail Applications
  • Airport Security
  • Mobile Solutions
  • Remote Wireless Monitoring of Matches
  • Data Protection & Human Rights

» Face Recognition

Dectel is a UK partner of Identix Corporation, developer of the world's leading face recognition software, FaceIt. Successfully integrated by Dectel into the town centre CCTV systems of Birmingham City Centre and Newham in East London, FaceIt automatically captures faces viewed by CCTV cameras and compares them with a database of stored facial images (such as known criminals). When a match is found, the control room operator is alerted and the appropriate action taken. The system is particularly useful in combating fraud by criminals using false identities or impersonation.

Automated face recognition of CCTV and photographic images has long been a goal of security managers and police forces. Up until now there have been few practical means of matching a face caught on CCTV with one contained on a watch list. Even when the images are recorded, someone would have the time-consuming task of looking through hundreds of printed photographs or hours of videotape recordings - a tedious job and therefore easily liable to human error.

FaceIt is a software engine based on advanced algorithms for automatically locating faces within complex scenes, tracking, cropping and identifying who they are. The system operates totally hands-off, continuously and in real time. Facial images from CCTV tape, live video or still photographs are transformed by FaceIt into a coded representation using an algorithm called Local Feature Analysis (LFA). The algorithm represents faces in terms of statistically pre-derived features from specific parts of the face. These features are the building blocks from which all facial images can be constructed.

FaceIt maps an individual's identity into a complex mathematical formula that can be matched and compared to a stored database of known or previously recorded faces. The system can be set up either to find a match/no-match or to locate the closest matches and rank them in order or resemblance to the target face. The database can be compiled from CCTV images, photographs or FaceIt's own online face capture facility.

There are many other applications for FaceIt, including access control and time-and-attendance systems. Also, the US Immigration and Naturalization Service is using FaceIt to improve transit for the many regular commuters who cross the USA/Mexico border.

FaceIt Technology

  • Face Detection detects single or multiple heads even in a complex scene
  • Tracking - follows faces over time
  • Segmentation - crops faces from background
  • Faceprint - generates digital code or internal template unique to an individual
  • Matching - one-to-one: authentication or one-to-many: verification

» Smart FACE id - Electronic Store Detective

Retail Crime, shoplifting and fraud, is increasing in the UK and with it there is a rising incidence of violence in particular against staff. CCTV together with Crime Reduction schemes and Partnerships have had a significant effect in reducing crime on a regional basis. Face Recognition Technology can assist in reducing crime even further and can also help to avert crime before it takes place.

Utilising Visionics FaceIt core technology, Dectel have developed Smart FACE id, a product specifically aimed at assisting store detectives in identifying known criminals at the earliest moment that they enter a store. Once identified an individual's activity can be monitored or a suspect can be requested to leave the store before a criminal act is committed.

The resulting benefits include a reduction in lost time attending court as a result of prosecutions, a reduced risk of violent confrontation with staff and an improvement in margin as stocks remain on the shelf to be sold.

In most cases existing CCTV cameras can be employed to capture the faces to be matched by the software. In some circumstances it may be necessary to add an additional camera to cover the most appropriate area.

Captured images are sent to a centralised computer for matching with a Watchlist of known criminals. Any matches that are automatically made are sent to the store detective via a hand held display for verification. If a positive match is confirmed additional information such as name and exclusion order details is available on the same screen. This on the spot availability of information can make the confrontation and request for the criminal to leave the premises a more low key (less inflammatory) event.

For retailers with more than one outlet, the database can be replicated to computers in other locations. This enables fraudulent activity across a store network to be tracked and identified in process. When used in conjunction with other communication systems such as e-mail and text messaging to mobile phones, the presence of a criminal in a given location, can be widely circulated in seconds.

In addition to detecting known criminals the Smart FACE id software provides many useful statistical reports. These include:

  • Number of Exclusion Orders generated in a period
  • Number of alarms generated in a period
  • Number of deterred entries within a period
  • Number of false alarms within a period

» Data Protection & Human Rights

Careful consideration is given to all products developed by Dectel using Face Recognition technology to ensure that application of the technology does not compromise compliance with the Data Protection Act 1998 and the Human Rights Act 1998.

Key Issues

  1. Any automatic match made by the software requires verification by a human operator.
  2. Any images captured by the software that are not automatically recognised as a match, are immediately deleted and not stored in any form of database.
  3. Any images that are captured and matched by the software that are not verified positive by a human observer are deleted and not stored in any form of database. An audit trail is also maintained of all human assessments of an automatic match.
  4. The technology is non-discriminating in regard to ethnicity, gender or age and cannot therefore be used to target individual groups.