Facial recognition system
Face recognition refers to the analysis of the expression of visible features in the area of the frontal head given by geometrical arrangement and texture properties of the surface.
Term discrimination
It is necessary to distinguish between the localization of a face in the image and the assignment of the face to a particular person. In the first case, it is checked whether and where a face can be seen in the second to who it is.
When it comes to the face recognition in the sense of knowing to what face it is, then we can distinguish two cases: If this is done by people, is in the English language of face perception is spoken, while face recognition by machine is referred to as face recognition.
Biological Face Recognition
The ability to detect and distinguish between faces ( face perception) is acquired by the human brain within the first months of life. It is of functions of the cerebrum, the occipital lobe accurate bound. A partial or complete absence of this ability is called prosopagnosia.
Technical face recognition
In a technical context, face recognition is one of the biometric methods. It is used in safety engineering, forensic and litigation, for the purpose of identification or verification ( authentication) of individuals. Typically, the technical, computer-based face recognition for access control to sensitive areas and search for duplicates in databases is used, for example in registers to avoid identity theft.
Decisive for the detection and digital representation of face images for interoperable purposes, in particular for use in electronic passports and criminology, is the international standard ISO / IEC 19794-5. His detailed specifications for image content and recording techniques aim at a high detection quality.
2D methods
Simple face detection methods use a two-dimensional (2D) geometric measurement of special features (eg, eyes, nose, mouth). Here their position, distance and location is determined each other. However, present methods require complex calculations, such as the best wavelet (for example, by means of Gabor transformation) or principal component analysis. The National Institute of Standards and Technology ( NIST) has repeatedly carried out comparative studies of various commercial and university procedures. The results show a significant increase in the recognition performance within about 10 years. Lag false rejection rate at a set false acceptance rate of 0.1 % in 1993 nor in practice disabled 79 % (ie, nearly four out of five people were at that time not recognized ), this error rate is now (as of mid-2006 ) of the most powerful method to only 1 % reduced (ie, about a hundred people is not recognized ). This rate is of the same order of magnitude as the current fingerprint or iris recognition methods and exceeds the capabilities of human face recognition.
3D method
In addition to the two-dimensional biometric face recognition, which is used for the detection of commercially available cameras, is currently developing a new branch, which sets collection ( eg by means of fringe projection ) of the face on the three-dimensional (3D). The additional information higher recognition accuracy and better poses independence and overcoming security are to be achieved. Test results of the NIST show that currently ( mid-2006 ), the 2D methods in terms of detection performance the 3D methods are still superior.
Application
The city of Rotterdam wants to test in October 2010 on a selected line for a year a face scanner. She wants to enforce the prohibitions imposed house and reduce violence and vandalism in the rail line.
In Germany, a project of the public broadcaster WDR was started during the Rhine cultural festival in Bonn in the fall of 2011, which allows face recognition of festival goers. This works on (high resolution ) photos of the festival-goers who are " tagged " using face recognition from Facebook. The faces are identified and linked to Facebook profiles. A similar project to have worked in Britain during the Glastonbury Festival already at 9,000 people.