Automatic recognition and detection of visual objects regardless of how they are depicted. That is, give computers the ability to recognise a horse, car, human etc whether in a photograph, a drawing, a painting, a toy etc. This is an EPSRC funded project EP/K015966/1) .../CRoDS/CRoDS.html
Non-Photorealistic Rendering from photographs and video. This work was all based on the proposition that seeing is necessary for drawing. It views artwork as salience maps that highlight elements of a scene that are important for visual understanding, and lead to automatic production of abstract and child-like art, including cartoonifying of real world video. It comprises a substantial body of work, much of which has been EPSRC funded.../NPR/Non_Photorealistic_Rendering.html
Automatic acquisition visually plausible models of complex outdoor phenomena. One example is trees: we can now capture 3D moving models of trees, and clone new trees from old. Known as the outdoor asset capture or OAK project (poetic license on the acronym), it enjoys EPSRC funding (EP/K02339X/1) , is joint with UCL and a team of 6 internationally leading companies.../OAK/OAK.html

My research is all based on the idea that artwork captures semantic meaning in photographic  imagery. This comes down to three main areas of interest.I have other interests too, notably in robotics and parsing mathematics texts.