• Rich Barlow

    Senior Writer

    Rich Barlow

    Rich Barlow is a senior writer at BU Today and Bostonia magazine. Perhaps the only native of Trenton, N.J., who will volunteer his birthplace without police interrogation, he graduated from Dartmouth College, spent 20 years as a small-town newspaper reporter, and is a former Boston Globe religion columnist, book reviewer, and occasional op-ed contributor. Profile

  • Alan Wong

    Executive Producer

    Alan Wong oversees a team of video producers who create video content for BU's online editorial publications and social media channels. He has produced more than 300 videos for Boston University, shuffling through a number of countries in the process: Australia, Argentina, Peru, Ireland, China, and Cambodia. He has also bored audiences in Atlanta and Boston giving talks on video for higher ed. Profile

Comments & Discussion

Boston University moderates comments to facilitate an informed, substantive, civil conversation. Abusive, profane, self-promotional, misleading, incoherent or off-topic comments will be rejected. Moderators are staffed during regular business hours (EST) and can only accept comments written in English. Statistics or facts must include a citation or a link to the citation.

There are 8 comments on Opening Movements

  1. It is undeniable that gesture recognition will have its future in many areas. However, I believe a sophisticated password generator or a finger print recognizer can build a far more secured and realistic system. Anyone could simply imitate gesture since the “input” of the “password” is so obvious.

        1. Yes indeed! Systems that use multiple biometrics to improve authentication performance are called multi-factor systems.

    1. Hello everyone,

      Thank you for your comments and interest in our work. We would like to respond to the questions that arose in these comments.

      First, it is important to note that gesture-based user authentication is not the same as gesture recognition. In gesture recognition, the goal is to identify the gesture being performed irrespective of who is performing it. Our goal, on the other hand, is to identify the *individual* performing the gesture and not the gesture itself.

      Next, in our envisaged application context, each user chooses a gesture that uniquely authenticates him or her. In this sense gestures are similar to signatures and passwords. The skeletal structure (relative lengths of limbs, etc.) is somewhat unique to each individual. In this sense a gesture is somewhat like a biometric, e.g., fingerprint, iris, voice, face — a physical property that is fairly unique for an individual. Now a gesture, like a signature, will typically be chosen by a user to be distinctive yet consistently reproducible by the user. This can be done, for example, by combining simple gestures in a sequence much like letters are combined to form words in language. In our preliminary experiments we have found that even when the gesture is fairly simple, it is non-trivial for someone watching it to reproduce it to the fidelity needed to trigger false authentication. It appears that like skeletal structure, the *gesture dynamics* too are somewhat unique to an individual even if he or she is performing someone else’s gesture with the express intent of accurately duplicating it.

      In summary, gestures, especially complex ones, are difficult to accurately imitate. On the other hand, a high-resolution photograph or a voice-recording can easily spoof a system that uses face or voice recognition. Fingerprints and iris scans are harder but they too can be “picked-up” and used for false authentication.

      Another key advantage of a gesture biometric over a conventional one such as a face, voice, fingerprint, or iris, is that it can be *revoked* when compromised. A gesture, if compromised, can be replaced by another. A face, fingerprint, iris, or voice cannot (at least not without significant pain to the user).

  2. Has anyone considered the issues involved when someone becomes injured and no longer has the same mobility or a range of motion they had when they made their “gesture” password? What then will they do? For example, they may have been playing a recreational sport and then dislocated their shoulder resulting in having to use a sling. It just so happens one of their gestures involves the rotation of that arm at the shoulder socket. Will they now be locked out until they heal?

    1. Excellent point. If such a situation should arise, the user can replace his/her old gesture with a new one or even repeat the old gesture under the new mobility constraints. As an analogy, imagine a person who needs to sign a bank-check but who has severely injured the fingers of his or her writing hand(s).

Post a comment.

Your email address will not be published. Required fields are marked *