Studienprojekte

Projekte sind ein essenzieller Bestandteil des Curriculums von Mobile Computing. Die Studierenden bekommen die Möglichkeit, das im Zuge ihres Studiums erworbene theoretische Wissen selbst praktisch umzusetzen. Ein sowohl für StudentInnen als auch für Lehrende immer wieder spannendes Unterrichtskonzept, in dem schon erfolgreiche Startups wie z.B. runtastic und Butleroy ihre Anfänge gefunden haben.

3D Face Segmentation Techniques for Mobile Devices

Zeitraum
Mar 2012 - Aug 2012
FH Studierende
Fabian Wenny, BSc
FH BetreuerIn
FH-Prof. Mag. DI Dr. Clemens Holzmann

The goal of the master thesis is the implementation of a robust 3D face segmentation technique on mobile devices. A main part is 3D image acquisition including stereo vision techniques on smartphones and the creation of a face database. Finally, a template based face segmentation algorithm is presented that provide images for face recognition systems on smartphones to give a securer, new possibility to authenticate mobile devices with the user's face.

Ziel

Since smartphones and tablets have the possibility to store private or sensitive data and additionally can be used as payment processor, mobile security has been come more and more important in the last years. One main problem in the mobile domain is a secure unlock technique for mobile devices. Currently there are a few possibilities to unlock the device as PIN, password or patterns. But all of them have the drawback, that shoulder surfing is used to get unlock information. Therefore, mobile device's overall security is not trusted that much at the moment.

One more robust way to unlock the smartphone is an authentication with face unlock using the frontal camera and a face recognition technology. Problems with illumination and unsecure 2D face recognition systems can be overcome with the additional usage of 3D information. The LG Optimus 3D is a smartphone with two build-in cameras and with the help of stereo vision techniques 3D information can be reconstructed and used for the face recognition. In order that 3D face recognition obtains good results, a correct face segmentation technique is prerequisite.

The main goal of this master thesis is the implementation of a robust 3D face segmentation technique applicably on mobile devices. Basically two types of 3D face segmentation techniques are presented: a nose-tip detection technique for 3D face segmentation with range images from a mobile phone and secondly a range template based face segmentation with average templates created from the u'smile face database images captured with the Microsoft Kinect. The segmented faces can be used in a face recognition system to authenticate mobile devices via face unlock in a new securer way.