Thank a Texas Engineer for the Visual Quality of Those Videos You’re Binge Watching
5 Questions with Professor Al Bovik
It’s not that common to hear about an engineering professor winning an Emmy. Unless they’re moonlighting as a TV star, the glitz of the Hollywood red carpet circuit isn’t the first place one might look for engineering expertise.
Then again, there aren’t many engineers like professor Al Bovik of UT Austin’s Cockrell School of Engineering. His work plays a key role in how those videos-on-demand and social media pictures make it from the camera and film set to televisions and smart displays around the globe.
In 2015, Bovik took home the Primetime Emmy Award for Outstanding Achievement in Engineering Development for his work on the development of video quality prediction models that are standard tools in the broadcast, satellite and video-on-demand internet industries. With a career spanning three decades, Bovik has published over 800 technical articles on perceptual picture and video processing and holds several U.S. patents related to this work.
This year, he is being honored with the world’s oldest (dating from 1878) and most prestigious award given in his field: The Progress Medal from the UK’s Royal Photographic Society. The Progress Medal is awarded in recognition of any invention, research, publication or other contribution that has resulted in an important advance in the scientific or technological development to photography or imaging in the widest sense.
We caught up with Bovik to learn more about how his picture quality measurement tools allow hundreds of millions of viewers worldwide to enjoy the best viewing experiences across platforms.
What kind of engineer wins Emmys?
The kind that is passionate about what they study and who likes working on interdisciplinary research problems. I developed a passion for translating visual neuroscience, which I learned on my own, into practical video engineering — a path I have followed for three decades.
All the algorithms we create in LIVE (our Laboratory for Image and Video Engineering at UT Austin) are based on mathematical models of real biological neural networks that process what we see. We also conduct large human studies, where we obtain many thousands (and sometimes millions) of human opinions of picture and video quality. We use these datasets to perfect the quality prediction power of our algorithms.
What is visual distortion and why does it matter?
Distortions are the defects that annoy people when they stream videos online or look at pictures or videos on social media. People see problems right away without even thinking about it like blur, noise, blocky-ness or stalls (when the video freezes with the dreaded ‘swirly’ in the center). Of course, they don’t like them. Our algorithms accurately measure our perception of video distortions by modeling the way the visual parts of our brain process such defects. Our algorithms were the first to be able to accurately predict whether a video is perceptually distorted, and its severity, and we can also identify the specific nature of the distortion.
We have injected the principles of visual neuroscience deeply into the internet, making it more efficient and able to deliver better visual media with fewer errors and reduced bandwidth consumption. This is important, since more than 75% of all internet traffic is composed of pictures and videos.
How does all this help inform the companies you work with?
We collaborate with many big players in the streaming video and social media space including Netflix, Facebook, YouTube, Amazon Prime Video and Oculus. My goals are two-fold: to give my students exposure to the most challenging problems facing some of the biggest players in the visual media space, and, by our work, to help our partner companies improve the visual experiences of the billions of people worldwide they provide services to.
How do you get students involved in your research?
I am constantly seeking fascinating and high-impact problems for the students in my lab to work on. When we create a new scientific theory, video quality model or algorithm, we publish it soon after on our website for anyone to use for free. This empowers other engineers and scientists around the world to create their own models and test them and compare them against ours. Everyone wins, especially the consumer. It is very gratifying that the algorithms developed in my lab help monitor and control more than half of all U.S. internet traffic.
For myself, I get to do what I love best — work with brilliant young minds to solve important problems that combine visual neuroscience with video engineering.
You are being honored with the most prestigious award anyone in your field can receive. Why is the Progress Medal significant to you?
Pictures and videos provide news, entertainment, socialization, health information and enable most forms of communication. They are especially important since we are such visual creatures; more than half of the human brain is implicated in vision. The Progress Medal honors accomplishments in all branches of photography and videography. It is the oldest and most prestigious award in the world for those doing work on the technical aspects of photography. I’m humbled, but this award recognizes the work done by my students as well. Recognition of this type gives us greater exposure, which will help me offer future students similar opportunities to shine.
It’s amazing and wonderful to be mentioned with Progress Medal winners from previous years, including such visionaries as the Lumières (inventors of cinema), George Eastman (founder of Kodak), Edwin Land (inventor of the instant camera and founder of Polaroid) and Tim Berners-Lee (inventor of the world-wide web over which most pictures and videos today are transported). These were people with a passion for science and engineering and for realizing it into technology that helps better peoples’ lives in meaningful ways.