Abstract

In today’s cameras, the transformation from the input image recorded on a camera sensor to the output picture is driven by the twin goals of first, representing in a physically plausible way the scene and second, producing a ‘good looking’ image. In this talk I will discuss specific contributions we have made in the Colour Lab at the University of East Anglia to camera processing pipelines including, the development of leading algorithms for white-balance, tonality and colour correction.

The pictures taken by tomorrow’s cameras will capture both what we ourselves saw but will also convey information that is invisible to the naked eye. As an example, pictures of scenes at distance or through haze often lack detail and yet the near-infra red images of the same scene are vivid and detail rich. Fusing this detail back into the RGB image image – using algorithms I will review – can produce preferred looking photographs. A second application of fusion processing, that I will discuss, is to make images that are, simultaneously, more accessible to colour-blind observers yet still look good to everyone else.

In the talk I will also touch on my experiences on commercialising our research.