4 June 2012. Gonzalo Bailador, postdoctoral researcher at the Facultad de Informática and member of the Biometrics, Biosignals and Security Group (GB2s), has developed two biometric techniques to verify the identity of mobile telephony users. One technique is to sign in the air holding the mobile phone in the other hand, and the other technique is to use the mobile phone camera to take a photo of one or both hands and authenticate the user from the shape of the fingers and hand.
Both techniques have been developed as software libraries for Android and iPhone mobile devices and can be integrated into any mobile application requiring an identity checking service.
Besides, the entire authentication process takes place inside the mobile device, and no biometric information has to be sent to a server. This assures the privacy of the user's biometric information.
In-air signature is a biometric technique that extracts signing hand acceleration features in each axis. These features are compared against a signature template that is created at an enrolment stage during which the user has to repeat the gesture three times.
The result of this comparison is used to verify the identity of the user who can then perform the protected operation. The robustness of this biometric technique relies on the uniqueness and fraud resistance of the user's chosen signature. This signature may match the user's handwritten signature or be a repeatable gesture.
In-air signature poses a bigger problem to forgers than other techniques based on traditional handwritten signatures, as they have to emulate how users hold and point devices while signing, and this depends on individual physical features, such as arm length, hand size or wrist turn.
In-air signature also has some other advantages over the handwritten signature, as no special device like a stylus is required, and users can sign quickly without even having to look at the screen.
The researchers have conducted several experiments to evaluate how precise the in-air signature technique is on mobile devices. To do this, they collected in-air signatures at different sites around the Universidad Politécnica de Madrid.
The 100 participants in database creation were recorded on video as they reproduced their handwritten signature in the air holding a mobile phone in their other hand. Later, other people studied these recordings and then tried to forge each of the signatures (six forgers per signature).
Based on this database, several in-air signature analysis algorithms were proposed and evaluated in order to minimize system error. The resulting forgery rate was close to 3%. This demonstrates the feasibility of this technique as a means of authentication, bearing in mind that the forger had a recording of the original signature.
The hand geometry technique is based on the fact that everybody has different hand geometry features, such as finger width or palm length. Based on three photos of the hand, the developed algorithms are capable of extracting these particular traits and generating a template for each person that will be used to identify the user. The characteristics actually used rely on the relative distances between the palm and four fingers, not including the thumb because of its mobility.
Compared with other hand geometry based methods, the major innovation of this technique is that the extracted characteristics are independent of hand rotation and camera distance. Consequently, it is very suitable for use in uncontrolled environments.
This technique is also operational in any lighting conditions, provided that the whole hand is uniformly illuminated. On the other hand, the segmentation stage is robust enough to be able to isolate the hand even against uncontrolled backgrounds like the street, home or office. Consequently, it is usable anywhere.
Different experiments on public and private databases of photos captured using mobile phone cameras yielded an error rate of less than 1%. This demonstrates hand geometry's feasibility as a biometric technique.
This line of research has been reported in several media (congresses, conferences or journals), and was funded by a public CENIT project (Segur@). The results of these algorithms merited second place in the 2011 International Hand Geometry Points Detection Competition (HGC2011), proposed by the Universidade do Porto, Portugal.
Smartphone proliferation, plus mobile broadband prices drops, have opened up the possibility of ubiquitously accessing a wide range of services.
The number of users accessing banking services from smartphones either to check account information or to complete electronic transactions has increased considerably. Some leading companies, like Google, propose mobile phone use as a means of payment using NFC (Near Field Communication) technology.
Being a multimedia device, mobile phones usually contain sensitive information about the user's person. This requires protection to prevent its use by unauthorized persons.
The biggest snag about using mobile phones to access such services is that terminals can get lost or stolen, and there is then a high risk them being used fraudulently by another person.
On this ground, a security measure has to be added to assure that only the mobile phone's genuine owner can perform these operations, and the two biometric techniques developed as part of this research are a response to this need.
University and business
The research has spawned a business project called BiomMo. BioMo was awarded a prize by the Universidad Politécnica de Madrid at the 9th UPM Business Start-Up Competition held this year.
The Biometrics, Biosignals and Security Group is also working on other biometric techniques such as iris, fingerprint, face, gait and smell recognition. The combination of these techniques with cryptography leads to cryptobiometry, a field of research aiming to generate cryptographic keys based on biometric features. The GB2S is embarking on several biosignals projects to detect human stress or diseases, such as cancer, based on body or breath odour.
Researcher Gonzalo Bailador explains the hand geometry technique. Photo: GB2s.
Gonzalo Bailador del Pozo, Carmen Sánchez Ávila, Javier Guerra Casanova, Alberto de Santos Sierra. Analysis of pattern recognition techniques for in-air signature biometrics. Pattern Recognition, 44, pp. 2468 - 2478 (UK): Elsevier, 01/01/2011. ISSN 0031-3203
Javier Guerra Casanova, Carmen Sánchez Ávila, Alberto de Santos Sierra, Gonzalo Bailador del Pozo. Authentication in mobile devices through hand gesture recognition. International Journal of information security (USA): Springer, 01/01/2011. ISSN 1615-5262
Alberto de Santos Sierra, Carmen Sánchez Ávila, Gonzalo Bailador del Pozo, Javier Guerra Casanova. Unconstrained and contactless hand geometry biometrics. Sensors, 11, pp. 10143 - 10164 (UK): MDPI AG,01/01/2011. ISSN 1424-8220
This item on other websites
INBIOSYS Biometría 20.60.2012
Mundo Insólito 05.06.2012
Keegy Argentina 04.06.2012
Tendencias Informáticas 04.06.2012