Some thoughts on Mobile Face Recognition (part 1)

Information access from smartphones and tablets has become mainstream both in business and personal environments over the last years. The use of these devices for accessing services like social networks, email or electronic commerce and banking has surpassed the access from traditional computers, turning mobile devices into essential tools in our everyday life. Mobility and ubiquity work are powerful tools for increasing efficiency and productivity in business (and also in personal life). However, without the proper usage, companies and users may
be exposed to security risks and threats.

Security in the access to information is one of the most important issues to consider in mobility scenarios. Passwords have been the usual mechanism for user authentication for many years. However, there are many usability and security concerns that compromise their effectiveness. People use simple passwords, they reuse them on different accounts and services, passwords can be shared and cracked, etc. The amount of different accounts and passwords we deal with these days contributes in making harder the proper usage and maintenance. As a result, we often see news and reports that alert of stolen accounts and passwords. This problem becomes critical in mobile devices, since they can be easily lost or stolen. Nevertheless, mobile devices can also become part of the solution, providing increased levels of security due to their new authentication options and capabilities.

The use of biometrics brings a more secure and convenient authentication method than traditional passwords. In the 2015 Biometrics Institute Industry Survey the use of biometrics for mobile access control has been established as the most significant development in the biometrics world over the last year. In addition, the survey points to other new applications for biometrics in mobile devices, such as mobile payments or law enforcement.

There are different biometric modalities that can be integrated in mobile devices: face, speaker, iris, fingerprint, etc. All of them have advantages and disadvantages, but one of the main benefits of face recognition (together with speaker recognition) is that, since smartphones already have integrated cameras, no additional hardware is required. Regardless of which biometric modality is used, for achieving a really effective system the following requirements must be accomplished:

  1. Usability: Ease of use is a key factor for achieving low false rejection rates.
  2. Security: It is important to avoid impostors to get access to the system (i.e. low false acceptance rate).
  3. Availability: The verification method should be usable anywhere and at any time.

Face recognition meets these requirements and brings a powerful biometric authentication solution for mobile devices since:

  1. It is easy to use and user friendly, since the user is already familiar with using the camera on the phone.
  2. Current face recognition systems achieve high recognition rates, suitable for secure authentication; and
  3. As stated before, face recognition does not need any additional hardware on the mobile devices. It takes advantage of the integrated camera so it is available in most smartphones.

However, there are some relevant issues for face recognition on mobile devices that remain unsolved or not enough studied. These concerns need to be addressed shortly for face recognition to be a leading contender in mobile device authentication. I will review some of these issues in future entries in this blog: liveness detection anti-spoofing methods, template protection, power consumption, availability under changing scenarios and adverse conditions or inter-device performance.

(The above text is extracted from this article, previously published in Image and Vision Computing Journal.)