By Bernd Mainzer, Atos

Screen shot 2013-07-30 at 17.07.25Electronic Machine Readable Travel Documents (eMRTDs), as defined in the ICAO Document 9303 and particularly ePassports, allow a high degree of border security and automation. This is one of the reasons why, in 2012, more than 104 countries issued ePassports and the number is increasing steadily. At the same time, the number of travelers at airports is growing at a record-breaking pace with current forecasts expecting 3.6 billion airline travelers worldwide by 2016.

Border and airport authorities face a huge challenge to improve traveller clearance and to enhance both border security and traveler experiences without increasing overall costs. The deployment of automatic border control systems (ABC, also called eGates) are expected to solve these issues. However, ABC systems are not limited to immigration at airports but can be implemented also at land and sea borders where eGates will gain more importance in the next years (e.g. through the eU Smart Border program).
The main goal of an automated border control system is to authenticate the electronic machine readable travel document or token by establishing that the traveller is indeed the rightful holder of the document or token. At the same time, the system queries border control records and then determines the eligibility of border crossing according to national, pre-defined rules.
One approach to establish a connection between a traveller and an eMRTD is by means of facial biometrics in eGates, based on a digital camera system requiring travelers to stop in front of it. This is nearly the same procedure as applying for a biometrically enabled eID document with live acquisition of the traveller’s facial data, also referred to as face image enrollment.
Thus, the traveller experience is very similar to the photo booth or photo studio experience. Indeed, the automated instructions are reminiscent of a photo cabin, like “stand at a fixed distance to the camera”, “don’t move your head”“keep looking into the camera”, “don’t roll, yaw, pitch your head or move your eyes sidewise”.
Whereas this is very important for facial enrollment, it has several disadvantages in verification scenarios of automated border control systems. Inexperienced travelers might not stop correctly in front of the camera nor might they look straight into the camera, which causes several face-capture retries and could decrease the expected traveller acceptance and throughput. A frequently incurring constraint is the diversifying height of travelers that needs appropriate solutions (e.g. height adjustment of the camera system, deployment of more than one camera).
State-of-the-art face recognition algorithms are challenged with requirements asking for better fRR/fAR metrics. This objective is easier to achieve if the reference image and the live images are captured under ideal photo studio conditions. The real challenge for the face verification components of an Atos eGate – including the algorithm – is to stay robust under non-ideal conditions. Significant progress has been made on this front, e.g. tolerance to face pose deviations or the robustness under varying light conditions. It is achieved without the necessity to force travelers to pose perfectly, or tedious iterations and interactivity when the applied face recognition system can cope automatically. The solution lies in balancing and optimizing relevant parameters, including light and camera settings.

Latest research and development in the area of facial recognition technology managed to overcome all mentioned disadvantages and paved the road to the next generation of Atos eGates, where intelligent eGate systems control the whole facial image verification process to make the usage of system
• easier,
• robust,
• more intuitive and
• more secure

The Atos eGate achieves all these objectives by a multi presentation face tracking system, optimized for score-driven face recognition with the help of an integrated sophisticated traveller guiding system. As soon as the traveller’s face is in the camera’s view field – even before the image stored in the ePassports chip is read out – Atos’ innovative face recognition technology automatically detects the face of the traveller and tracks it until clearance. This new approach was put into place, with a bank of four parallel Atos eGates, at Sofia airport in the summer of 2012, and makes Sofia airport one of the first adopters worldwide of next generation eGates provided by Atos. These eGates fulfill all legal requirements and follow border control process recommendations defined by the eU legislation and Frontex as well as local border authority regulations. A high level of security using automated border control systems is not only achieved with biometric verification of travelers, but relies heavily on a combination of physical security, fraud protection (e.g. altered passports, spoof protection) and background checks in national and international databases.
Security and safety considerations include the reliable design of physical barriers to avoid jump-overs and crawlunders. This can be achieved by assuring that the sensors cannot be manipulated (e.g. by a cover). The eGate shall provide measurements against tailgating and piggybacking and should incorporate left item detection down to the size of passports.
Spoof detection gains importance and becomes a standard requirement for ABC systems. The development that has already started with the request for facial 2D spoof detection will lead to a fight against impostor attacks with more and more sophisti- cated methods. High levels of border security clearances are one objective for eGates but at the same time, a preferably high traveller throughput should be possible. The overall clearance time is influenced by ePassport reading, biometric verification performance, eGate mechanics (e.g. by opening and closing doors) and, most importantly, by the traveller. In order to let the traveller know how to behave and what to do, an easy-to-understand guiding system has to be established. This kind of information is not only required right at the eGate but is necessary before the traveller has reached the eGate. Possible delivery methods include “how to use eGates” videos, animations and flyers, amongst others.
emerging new technologies, like the new generation of RFID chips and operating systems, as well as communication protocols (e.g. VHBR), once adopted in eGates, can help to increase the traveller throughput.
The Atos next generation eGate solution is based on Atos’ partner Gunnebo® ImmSec gate, a specifically designed double gate interlock solution used for the automated validation of travelers at immigration checkpoints and border crossings. The slim and ergonomic design allows the ImmSec to be installed in single or multiple lane configurations, providing an open feel to the traveller without compromise to security. It comes as a one- and bi-directional two-door gate running the Atos next generation eGate software solution with multi-modal biometrics (face and fingerprint). It is equipped with an electronic document reader (e.g. passports, ID cards or VISA), a system for single person detection (SPD), anti piggy backing (APB), anti tail gaiting (ATG) and left item detection (LID), and a fingerprint reader. Atos next generation eGate software solution comes with an administration-, configuration and supervision-application, the later allowing one single border guard to supervise several (configurable) eGates in parallel overseeing all transit steps visualized on the screen.
The face biometrics performance is based on the market leading face recognition technology of Cognitec Systems GmbH together with the camera system of partner MoDI Modular Digits GmbH with its Adomo® technology to capture high resolution images. The combination of Atos next generation eGate software, Gunnebo, Cognitec and MoDI modules provides a uniquely fast and secure eGate solution.

This article was first published in the Vault Magazine. Please refer to the pdf version for footnotes and sources.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: