• 3D cameras
• camera-projector systems
• depth cameras
• gesture based human-computer recognition
• gesture recognition
• Kinect
• mobile computing
• text tracking

In this thesis we present a portable camera-projector system controlled by a gestural human - computer interface. The system identifes research
artefacts projecting related information directly back into the
research space. Our main assumption is that the objects present text
data on their surface and that they are in the device's eld of view.
Traditional camera projector systems for text recognition employ only
standard RGB cameras. These type of devices can be employed for a
variety of applications but little attention has been paid to their application
in the task of recognising texts in natural scene images. Previous
work has been centred on the location of text in whole images, without
any real interaction and requiring good resolution across the whole
image. The novelty of our approach lies in the using a depth camera
to assist the system in the recognition of artefacts of interest, making
text tracking interactive and therefore reducing the resolution demands.
Design concepts, a prototype implementation and experimental results
that show the validity of our approach are presented. Furthermore, in
order to create a portable device, we undertake a preliminary investigation
on the feasibility of jointly employing depth cameras with small
sized computer units, which to the best of our knowledge has never
been done before.