||Most images and videos are capture, stored, manipulated and displayed with limited fidelity and accuracy. Capture devices typically produce data that is limited and quantised in time, space, gamut and dynamic range. Subsequently, nearly all processing, transmission and display is achieved with limited fidelity. As a result, effective communication by means of visual media is sub-optimal, the immersive qualities of displays can be improved, and image processing and computer vision algorithms are required to make sense of data that has already lost information that cannot be recovered.
The aim of our research is to develop algorithms and techniques to remove these limitations, and to convey a sense of visual realism. We investigate color-accurate image processing, and aim to deliver techniques for the calibrated capture, transmission and display of images and video. In cases where physical accuracy is impractical or impossible, for instance in certain inverse problems, we aim to develop perceptually plausible algorithms and demonstrate their effectiveness by means of visual psychophysics.
In 2013 Erik Reinhard joined Technicolor Research and Innovation in Rennes, France, as a Principal Scientist.