Project (M.A.) - Simon Fraser University, 2005. / Project (Dept. of Economics) / Simon Fraser University. Also issued in digital format and available on the World Wide Web.

Thesis (M. Phil.)--Hong Kong University of Science and Technology, 2004. / Includes bibliographical references (leaves 71-76). Also available in electronic version. Access restricted to campus users.

Vision based technology such as motion detection has long been limited to the domain of powerful processor intensive systems such as desktop PCs and specialist hardware solutions. With the advent of much faster mobile phone processors and memory, a plethora of feature rich software and hardware is being deployed onto the mobile platform, most notably onto high powered devices called smart phones. Interaction interfaces such as touchscreens allow for improved usability but obscure the phone’s screen. Since the majority of smart phones are equipped with cameras, it has become feasible to combine their powerful processors, large memory capacity and the camera to support new ways of interacting with the phone which do not obscure the screen. However, it is not clear whether or not these processor intensive visual interactions can in fact be run at an acceptable speed on current mobile handsets or whether they will offer the user a better experience than the current number pad and direction keys present on the majority of mobile phones. A vision based finger interaction technique is proposed which uses the back of device camera to track the user’s finger. This allows the user to interact with the mobile phone with mouse based movements, gestures and steering based interactions. A simple colour thresholding algorithm was implemented in Java, Python and C++. Various benchmarks and tests conducted on a Nokia N95 smart phone revealed that on current hardware and with current programming environments only native C++ yields results plausible for real time interactions (a key requirement for vision based interactions). It is also shown that different lighting levels and background environments affects the accuracy of the system with background and finger contrast playing a large role. Finally a user study was conducted to ascertain the overall user’s satisfaction between keypad interactions and the finger interaction techniques concluding that the new finger interaction technique is well suited to steering based interactions and in time, mouse style movements. Simple navigation is better suited to the directional keypad.