Faculty Focus: Using science to push the boundaries of photography

Prof. Wolfgang Heidrich

KAUST's Prof. Wolfgang Heidrich, Director of the Visual Computing Center and Professor in Computer Science, was recently awarded the Humboldt Research Award in recognition of his lifetime achievements to date, particularly within the research area of Computational Photography and Displays. This is an area of scientific research in which researchers are attempting to push the boundaries of photography with the aid of computers and software.

The award comes with a prize of €60,000 and a research stay in Germany. As the host institution, Heidrich chose the Max-Planck Institute for Informatics and the Cluster of Excellence on "Multimodal Computing and Interaction", both located on campus at Saarland University in Germany.

We sat down with Prof. Heidrich to discuss his award win, his research and his goals, including his future ambitions here at KAUST.

Professor Heidrich, can you describe the Humboldt Research Award in more detail?

The Humboldt foundation is a German organization which funds international collaboration. They have a host of programs for bringing foreign Ph.D. students and foreign postdocs to Germany, along with hosting faculty and sending German faculty and postdoc students aboard. The Humboldt Research Award is one of the higher profile faculty awards. It is for people living outside of Germany all around the world; it's across all disciplines and comes with €60,000 prize money, but mainly its aim is to try to bring high profile people into the country to network with German researchers.

You have chosen to carry out your research at the Max-Planck-Institute for Informatics in Saarbrücken?

Yes, over the course of the next three years I expect to spend roughly 6 months in Germany talking to various faculty members and hopefully doing some research projects together. In fact, this has almost already started in parallel with my appointment here at KAUST. I feel it's going to seamlessly integrate with the plans I have here. My official host for my stay in Germany is Hans-Peter Seidel who is a director of the Max-Planck Institute. I will visit and spend a time at the Max-Planck Institute but I will also use my time to visit various other groups and universities within Germany. From my time as a Ph.D. student back home, I know a lot of faculty members in my research area in the University of Bonn and TU Munich, so I'm going to use this award to reconnect with some people in my field.

What brought you to KAUST and when did you join?

I have been here since February 2014. This place is such a vast, fast moving and growing place that for the short nine months I've been here I feel like I have been here for a lot longer. There are two standout reasons I chose to come to KAUST. The first one being that I had a full professor tenured appointment at the University of British Columbia and I had done that for a couple of years, so I was looking for some new challenges. Becoming the center director here came at the right time and it was a nice fit for me.

The second reason was the people here and the research opportunities. In terms of facilities alone, KAUST is really the best place that I've seen. I mean honestly the research infrastructure here is phenomenal. I don't know any other university that comes close. What I like most about KAUST is that you have so many facilities and labs that are accessible to everybody, even to people you think might not need them. In fact that's what happened me with the nanofabrication lab. I never really considered myself as someone who needs a nanofabrication lab, but then you come here and see what is possible, the opportunities for training that my students and postdocs can receive to build custom diffraction grading's for some optics projects, and you cannot help but avail yourself of the facilities at hand. When you see what is here at KAUST there is a massive scope for personal development, what were ideas before I have been actually able to realize them here at KAUST that was not possible in other places where I have worked.

Have you benefited from the opportunity of personal development and growth here?

Yes, I have expanded my original research area of computer graphics and computer vision into for example optics. The team I can assemble here, including, students, postdocs, and research staff of various different backgrounds, as well as the actual nanofabrication lab and other core labs here are really helping me have a research agenda that just wouldn't be possible anywhere else.

Can you outline your research here, what is the focus of your research?

My research here is computational imaging and photography, we are aiming to co-design imaging hardware and algorithms for cameras and better, more functional displays. Traditionally what has happened in camera design is that there were people worried about designing image sensors such as, complementary metal-oxide-semiconductor chips (CMOS). Then there were others worried about making the optics as good as possible, and finally those who developed algorithms to correct deficiencies or artefacts that are left in the image to give you a good image at the end of the process.

As it turns out there is a much more powerful way to go about it where you think about the algorithms, the electronics, and the optics as variables you can control at the same time, to give you the best possible overall performance. For example, instead of having more and more complicated camera lenses with more and more individual lens elements, you can think about what you can do with fewer lens elements, by shifting some of the design effort from optics into algorithms. This in turn will allow you to make your optics thinner and also cheaper and more lightweight. For example, if you can change the optical design of your camera in such a way that it becomes thinner you then can start to make cellphones thinner.

However then the raw image quality that come out is going to be worse, but if you can control in which way it is worse, then to some degree you can design optics in such a way that the artefacts have a certain structure. Some structures are relatively easy to correct with software, while other artifacts are very hard to correct. Basically the goal is to co-design the software and the hardware in such a way that you get the best overall performance and ensure that the hardware doesn't create artifacts that are impossible to correct for in the software. A similar principle of hardware-software co-design applies not only to cameras, but also to display technologies. Were I have worked on topics such as high contrast (high dynamic range) as well as auto-stereoscopic displays (i.e. glasses-free 3D displays) over the years.

I read one of your quotes where you say that you have developed a computing method in which you can display high contrast displays on monitor screens as humans would perceive them in reality. Can you expand on that?

It is a whole research area for me -- this high contrast imaging. Basically it started with the display hardware, which was actually the first example of a "computational display" in the sense of having been based on hardware-software co-design. This lead to a technology called "local dimming" displays, that was commercialized by a startup and eventually acquired by Dolby, and is now found in most flatpanel TVs that you can buy. From this work on the display hardware, we the expanded to also look at how you take video streams that don't have such a high contrast range and try infer what the high contrast image range might be. So for example, if I take an image out the window here now, you will see all the highlights in the water and they will just be completely saturated in the normal image I take. But again using some assumptions about natural scenes you can actually infer which points should come across as brighter than white over say that of a normal photograph. You can then actually restore the image to some degree, of course it is not going to be physically accurate but visually it will be plausible.

What led you down this path? What fueled your interest in your field?

I've always been interested in computation and what I like about computational imaging and display is that you can combine the computational aspect with something else that is very visual. You can see the immediate feedback in terms of a better image quality. You do need a lot of theory and math and abstract skills but the output is not abstract at all, it is the image itself.

By David Murphy, KAUST News