This year have seen the 25th edition of the Color Imaging Conference (CIC)) taking place in Lillehammer (Norway) after San Diego (California).

This time again I could be part of the committee as a co-chairman for the Short Course session. Session where conference attendees or not attendees can register and take courses (you need to pay for each course you want to follow) on carefully selected topics. Basis of color science courses are of course available but also several advance courses on different topics given by field specialists (e.g courses on hdr, color differences and other colorful themes). The instructors are usually coming from academia and/or industry - when their industry allows them to talk - and each course is first submitted, reviewed and/or not accepted.

Back to Norway

I have been visiting Norway regularly after I left Gjovik in 2008 and its colorlab (a few km South of Lillehammer) where I did my PhD on multispectral color reproduction. It was something to come back in the lab, seeing old colleagues, seeing that the color lab did grow nicely and was heavily represented at the conference.

General overview

Over the years you appreciate the conference with different perspectives. Now that I know a bit more about my field - color science - I can witness that each sub-field is evolving slowly, appreciate tiny changes, observe the increase of general knowledge. There is always something.

Human perception and colour constancy

Keynotes, first and second have been pretty interesting. The first one given by Apple - almost did coincide with the Apple event in California - was impressive and disturbing in the same time. Impressive because of the amount of advanced technology they are able to squeeze into a smartphone. Disturbing by the non research dissemination they are doing. They have a great team of researchers and engineers but all the knowledge stay at work, not sure it will make the people smarter at the end. I’m sure they are not the only company doing that, but they are so big the result of such behavior has an impact.

Second keynote was entitled Twenty-five Years of Colour Constancy given by Anya Hurlbert from Newcastle University (UK). Needless to say it was pretty entertaining, a the typical presentation that make you feel smarter at the end. Colour constancy is a fascinating phenomena which can be understood as white balancing done by the brain. We have not all the same automatic white balance algorithm embedded with us, this resulting endless discussion on object color, remember the blue-black/yellow-white dress-gate a few month ago?

Subjective evaluation, image metrics and database

There are so many metrics it’s becoming difficult to follow on that topic. To have a idea have a look at the Tampere Image Database TID203. Each of them is trying to mimic how human perception is working, allowing to evaluate image difference, predict how image quality will be perceived when a new algorithm is applied to images.

A solution is to follow an experimental workflow which integrate feedback from standard real observers. A panel of test persons is asked to look at images, original and modified and to compare them. Notion of just noticeable difference (JND) is of course very important here. In that case subjective evaluation is often chosen, but here are well they is plenty of existing experimental workflows to choose from.

Human vs bird viewing system

Going to the same conference each year sometimes shows you how alone are researchers. Most of them are spending their life on one topic and they become hyper specialist. A conference is the occasion for them to leave their laboratory, office and to share their discoveries (in that case other color scientists).

A peculiar article tilted Super Vision Model: What’s Peking Robin Seeing? by researcher Hiroaki Kotera from Kotera Imaging Laboratory, Chiba, Japan was pretty cool. This research work looked at birds (the Peking Robin) and their visual system: they are tetra-chromate RGBU when we are tri-chromate RGB. Question being how these animals see colors? Four sensors and tiny brain Vs us three sensors and bigger brain that allows to post-process the signal and compensate for a lower captor resolution or how nature optimizes living organism for dedicated tasks within an ecosystem. The non-visible sensor allowing this bird to evaluate more precisely the quality of food (e.g. is this fruit already rotten or not?). A parallel can be done with computational photography when the algorithms and post-processing present in smartphone can compensate for their lower quality hardware (to some extend) compare to high-end dslr.

Workshop session

I could attend only two of the three parallel workshop sessions. interesting on many points. Both workshops W2: VISUAL PERCEPTION AND EMERGING TECHNOLOGIES IN CINEMA: PERSPECTIVES FROM ACADEMIA AND THE INDUSTRY and W3: CULTURAL HERITAGE DIGITIZATION: CHALLENGES AND OPPORTUNITIES shared cutting edge used of technologies. More important they allow to have discussions between professionals from different entities without revealing company secrets.

La suite

Hope to be part of next year edition in 2018. For now the next destination hasn’t been revealed, but according to recent tradition it should return to North America.