About UCT Computer Vision Group

Much of the visual information around us today is represented and processed digitally. Digital Image Processing (DIP) is ubiquitous, being present in everything from our televisions and our mobile phones to our digital cameras. Beyond these consumer applications, it is used in many other sectors too, from medical imaging and diagnosis to molecular biology robotics and remote sensing. Also termed ‘computer vision’, DIP essentially involves how computers or machines see the world. At the UCT Department of Electrical Engineering, postgraduate students in the DIP group focus on research in this area.

Research currently pursued by the DIP group is multi-view 3D reconstructions and the subsequent use of these for classification and optical sensing. The group also has expertise in medical imaging and computer tomography.

Research projects performed by the DIP group include the following:

• Shape from silhouettes (object shape estimation from multiple silhouette views);
• Particle analysis (estimating geometrical properties of particles for example 3D size and shape);
• Multiple camera person tracking (tracking humans in an environment from multiple overlapping cameras);
• Multiple view geometry reconstruction (3D reconstructions from multiple views of an object);
• Froth and rock analysis (size and shape analysis of froth and rocks for mineral processing);
• Medical imaging and tomography (performance characterisation and enhancement of the Lodox scanning X-ray machine).

The group maintains a strong emphasis on industrial applications and receives funding from De Beers, Anglo Platinum, Rio Tinto, Lodox Systems, the National Research Foundation and the Department of Trade and Industry.

Those who emerge from their specialisation at postgraduate level can choose from a number of careers in the DIP field across a range of sectors, where the demand for skills is high, from telecommunications and the automotive industry to medical imaging and industrial research.