Tuesday 13 December 2011

Multi-Spectral Connections

One of Alejandro's many test images, looking at the characteristics of multi-spectral document imaging. Used with Permission.

Another interdisciplinary research project I am currently working on is with Adam Gibson, from The Department of Medical Physics and Bioengineering at UCL. Adam and I started at UCL at the same time, and met on the teaching training course that all new staff have to do. At one point he remarked that he did some work in multi-spectral imaging, to which I replied "oh yeah, we look at documents sometimes using that technique" - although I described our usually laissez faire approach of sticking things under the filtered lens and seeing what you can see, which turns out to differ greatly from tried and tested benchmarked multi-spectral methods used in medical physics to, for example, measure blood flow through the body.

5 years later, Adam won a very prestigious EPSRC Challenging Engineering grant on "Intelligent image acquisition and analysis". He got in touch to wonder if we could use a small slice of money to investigate the potential crossovers between the medical physics methods of multi spectral analysis, and the approaches we used in document imaging analysis. Can the robust, tested, techniques used in medical physics be used in document analysis? Can we benchmark the process of multi-spectral imaging of documentary material?

Our PhD student working on this is Alejandro Giacometti. Alejandro has a background in computer science, and came highly recommended from the MA in Humanities Computing under the supervision of Stan Ruecker at the University of Alberta. Alejandro is really well placed to carry out this research, having the technical background as well as appreciating the Humanities angle. Simon Mahony from UCLDH has also joined us on the team, and he brings his knowledge and expertise in Digital Classics.

Alejandro is now almost half way through his thesis work - and as well as being really eye opening, this project is turning out to be so much fun. We are now in the phase of starting to test our hypotheses on real world examples, and building up our practical expertise in multi-spectral document imaging. It also turns out that we have a lot in common with both the work of Tim Weyrich from UCL Department of Computer Science (who I also jointly supervise a student with) who uses multi-spectral imaging to model skin surfaces, and Stuart Robson from the UCL Department of Civil, Environmental and Geomatic Engineering (who I also jointly supervise two students with, but shall talk about them at some later point) who is interested in a huge variety of image capture techniques both for industry and heritage. We're talking between teams, departments, and disciplines, now, and learning a lot from each other, while cooking up plans for future work.

I feel very lucky to work in an institution such as UCL which has such diverse expertise - but also such interested colleagues, willing to work together across disciplinary boundaries. Its also great to have such an opportunity for PhD study, which could potentially contribute to many fields.

2 comments:

Jeff said...

You might be interested to learn of a multi-spectral imaging project on David Livingstone's unpublished letters and diaries, by Adrian Wisnicki and a team of imaging scientists.

http://www.nytimes.com/2011/11/08/science/restored-livingstones-fading-notes-from-africa.html?_r=2&ref=science

http://www.washingtonpost.com/lifestyle/style/dr-livingstones-diary-on-19th-century-africa-now-uncensored/2011/10/31/gIQAUsB2aM_story.html?hpid=z9

http://www.bbc.co.uk/programmes/b010dk1z

Melissa said...

Yes! thanks, we had heard of this. Its the type of thing we are pursuing, too - starting to look at real use cases. It will be great to get the maths all square, but actually, the interesting thing for Humanists is uncovering new documentary evidence!