Abstract
Modern mobile devices have become very powerful. All kinds of sensitive data are
stored and accessed through these devices. Mobile banking, m-commerce and web- browsing are some of
the activities one performs with smart phones or tablets. The security of the data stored and
accessed through mobile devices (like emails and private documents) is crucial.
Password protection has proved to be weak for mobile phone protection. Biometrics security has been
developed and is used more and more on mobile devices. The use of facial recognition for
authentication on mobile devices is an extensively researched area. Issues that are usually
encountered are the limitations of the mobile devices in terms of limited memory, limited
processing power and poor image quality, although these are improving quite rapidly.
The purpose of this research is to explore the different solutions to the problem of improving
mobile device security using facial recognition technology. The diverse solutions proposed in the
literature were investigated to determine what solutions had been proposed and implemented and to
determine which ones gave the best results. This was done in such a way that the chosen solution
was compatible with the limited hardware and memory capabilities of these devices.
In this study, many image processing algorithms from the available literature were explored.
Feature-based facial recognition systems were found to be computationally less expensive and
relatively simple to implement.
An algorithm based on geometric measurements of the face and the Golden Ratio was proposed and
implemented on an Android platform. The application was tested
on the Samsung Galaxy Tablet 10.1.
M.Ing. (Electrical and Electronic Engineering)