Probably, we have all seen CSI and other forensics based dramas where some indelible piece of evidence leads to a serious criminal being caught. Often the piece of evidence is not visible to the naked eye it may only become apparent if viewed under, for example, Ultra Violet Light. Like these popular fictional dramas this project is concerned with making the invisible visible.

However, we are interested in the problem of augmenting the visible information (with detail that is invisible to the naked eye). As an example, the image taken with a thermal camera (which measures heat) is bright wherever people appear in a scene. If we fuse the thermal image with a normal colour photo (of the same scene), the position of people ‘pops out'[1]. Making people ‘pop out' is tremendously useful in applications such as surveillance and tracking. However, ideally, the fused image should still look, more or less, like a normal photo. Otherwise the image could look weird and any weirdness may be more difficult to interpret the image (in technical terms, by augmenting the visible image we do not wish to add to an observer's cognitive load[2]).

In the proposed PhD project, the successful applicant will develop an imaging system where visible imagery is fused with the output of an Near Infra-Red Camera. The aim is, though image fusion, to provide a modified augmented colour image with respect to which it is ‘easier' for people and machines to pinpoint the location of any  faces that may be present. To test face detection in people, a psychophysical detection experiment will be carried out (we would like to test the hypothesis that  observers will be able to find faces more quickly in the augmented image). The machine-intelligent system will, as a first step, test recognition by incorporating image fusion into a standard face classifier[3].

There will be scope in this project to apply the developed techniques to other domains including forensic imaging.The successful applicant will also have the opportunity to work with Spectral Edge Ltd (a recent UEA spin out company in the area of image fusion).

Jonson, M.J. and  Bacjsy, P. (2008), "Integration of thermal and visible imagery for robust foreground detection in tele-immersive spaces," Information Fusion, 1-8.

Matsukura, M., Brockmole, J. R., Boot, W. R, & Henderson, J. M. (2011). Oculomotor capture during real-world scene viewing depends on cognitive load. Vision Research, 51, 546-552.

Viola, P.A., Jones M.J. (2004), ":Robust Real-Time Face Detection," International Journal of Computer Vision 57(2): 137-154.