Iris recognition
The iris recognition is a method of biometrics for the purpose of authentication or identification of individuals. For this purpose, images of the iris (rainbow skin) of the eye are recorded with special cameras , the characteristic features of the respective iris are identified using algorithmic processes, converted into a set of numerical values ( feature vector , "template") and for recognition by a classification algorithm such as . B. a neural network is saved or compared with one or more already saved templates.
Development history
The original concept of using iris images for biometric recognition was developed by Flom and Safir in 1987 and a patent pending was applied for. The expiry of the patent in 2006 has since led to increased research efforts worldwide.
The currently most widespread method and template (as of April 2007) in commercial use is the iris code based on algorithms by the mathematician John Daugman .
The US company EyeLock has been offering the first iris scanner with USB connection that is compatible with the U2F protocol of the FIDO alliance since January 2015 .
properties
Commercial recognition methods detect around 260 individual optical features of the iris. These characteristics develop from a random, morphogenetic process in the first months of a person's life and remain largely unchanged over the rest of their life. Identical twins do not have an identical iris structure either. The outstanding property of iris recognition in practical use is its extremely low number of false positive comparison results ("false matches") compared to other biometric methods. that is, the likelihood of mistaking an iris code for that of another person's eye is close to zero. This means that iris recognition is suitable as a reliable identification method even in large databases with millions of personal data records, as well as for identification in access control situations without a primary identifier, i.e. without ID cards or RFID tags.
False negative results ("false non-matches"), in other words cases of not recognizing a person actually recorded, can arise, especially in unfavorable recording conditions of the eye, if the iris is due to the edges of glasses, reflections on glasses or the mostly East Asians ( or Far East Asians) typical narrow eyelids are not sufficiently visible.
Another characteristic is the low need for computing resources for the iris comparison. Iris recognition and recording is therefore particularly suitable for mobile use in PDA- sized devices.
Applications
Examples of application are the iris recognition used by the UNHCR in the repatriation of Afghan citizens and mostly Syrian refugees in Jordan and neighboring countries, in border police entry controls in the United Arab Emirates or in the control of protected areas in Iraq by the US Army. In all these cases, iris recognition is intended to ensure that none of the controlled persons appears under different or false identities. Other applications include access control to high-security areas and automated border controls (e.g. Frankfurt , Schiphol , Heathrow airports ).
The currently (as of 2011) world's largest project with biometric iris recognition is the Indian " Aadhaar " program. It is used to record all around 1.2 billion people residing in the country so that they can be assigned a unique personal identification number for government and legal purposes. A combination of iris and fingerprint recognition should enable the establishment and avoidance of multiple or false identities.
During the first mass application in immigration control in the United Arab Emirates from 2002, there were attempts by illegal travelers to avoid identification by means of iris scanning. With the help of medical eye drops, the pupil expands and the iris contracts, so that recognition is temporarily prevented. The system was then modified so that an enlarged iris is automatically recognized as such and the iris scan is repeated on the person after a few hours.
Together with face and fingerprint recognition, iris recognition is one of the biometric forms provided by the ICAO for use in electronic passports ( ePassport ). In order to guarantee the worldwide, manufacturer-independent interoperability of the data, the ISO / IEC 19794-6 standard “Exchange format based on iris images” specifies the requirements for iris image recording and storage.
Iris recognition is only suitable to a limited extent for forensic purposes, as the iris structures disintegrate just a few minutes after a person dies.
In cell phones, iris recognition is used to unlock the phone instead of a PIN code or fingerprint. This technology was first used in the Fujitsu ARROWS NX F-04G. The mobile phones of the Lumia 950 series from Microsoft also use this authentication technology. The Chaos Computer Club demonstrated in May 2017 how easily the iris recognition can be overcome on the Samsung Galaxy S8 .
Data protection issues
Today's commercial devices for iris recognition have in common that the distance between sensor (camera) and eye is limited to around 0.2 to 1 meter and recognition requires cooperation by the user - he must actively look in the direction of the camera - but what can also be provoked voluntarily by hiding the sensor in an object that functions as an " eye-catcher ", for example an attractive advertising poster or an information board. In the feature film Minority Report , a fictional future is described in which iris recognition systems automatically identify passers-by at a distance of several meters. The development of such systems is currently (as of April 2007) in preparation in the form of feasibility studies and prototypes. It is feared that totalitarian states or organized crime could misuse the identification capabilities of iris recognition for dishonest purposes. Therefore, there are considerations regarding the need to put the spread of such devices under legal control.
See also
swell
- ↑ L. Flom and A. Safir: Iris recognition system , US Patent 4,641,349, 1987
- ↑ EyeLock's myris Is First and Only Iris Authenticator for New FIDO Open Industry Standard , EyeLock, accessed January 5, 2015
- ↑ NIST Report 2005 (English)
- ↑ NIST Report 2006 ( Memento of the original dated July 6, 2007 in the Internet Archive ) Info: The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to the instructions and then remove this notice. of the Face Recognition Vendor Test (English)
- ↑ Christina zur Nedden, Ariana Dongus: Biometrics: Tested on millions of involuntary people . In: The time . December 17, 2017, ISSN 0044-2070 ( zeit.de [accessed December 18, 2017]).
- ↑ Unique Identification Authority of India Homepage of the Indian UIDAI authority
- ↑ Iris scanner blocks 62,000 illegals Abu Dhabi Police, May 3, 2006
- ↑ Fujitsu Releases ARROWS NX F-04G - Fujitsu Global. In: www.fujitsu.com. Retrieved September 15, 2016 .
- ↑ Chaos Computer Club hacks the iris recognition of the Samsung Galaxy S8. Retrieved May 23, 2017 .
- ↑ Iris recognition from a distance
- ↑ Press release on export control considerations
Web links
- Lecture on the function of iris recognition at the Humboldt University Berlin (PDF file; 680 kB)
- Total surveillance starts in Leon, Mexico ( Memento from August 22, 2010 in the Internet Archive )