Thomas S. Huang


Signal processing and analysis. In terms of signals, I mainly work with images and video and, more recently, speech and acoustic signals. In terms of the processing, there are three main parts of processing. One is compression, the second is enhancement or improving the quality, and the third is pattern recognition.

Written by

Thomas S Huang
Thomas S Huang

Q: What is your area of expertise?
A: Signal processing and analysis. In terms of signals, I mainly work with images and video and, more recently, speech and acoustic signals. In terms of the processing, there are three main parts of processing. One is compression, the second is enhancement or improving the quality, and the third is pattern recognition.

Q: Give me a brief synopsis of your education and career.
A: I got my BS from National Taiwan University and my MS and doctor of science from MIT. I taught at MIT for 10 years, Purdue for seven, and then since 1980 I’ve been at Illinois.

Q: You’ve attended and taught at MIT, and you’ve taught at Purdue. What made you come to Illinois?
A: First of all, if you stay at one place for too long it tends to get stale, so the intention originally was to move every 10 years or so. I taught at MIT for 10 years and then moved to Purdue. I stayed there for seven years and then moved to Illinois. And actually after 10 years at Illinois, I was thinking about moving, except then we had the Beckman Institute. It has a completely new kind of environment, which I like so much that I haven’t thought about leaving. But if not for the Beckman Institute, I would have probably moved after the 10 years here.

Q: What do you do at Beckman that has led you to stay here so long?
A: The main thing with Beckman is that it tries to encourage interdisciplinary research. So there are people from many, many different departments, and it’s very easy to collaborate. Now the world has become so complex, so solving complex problems usually needs expertise from different areas, and the physical proximity at Beckman makes collaboration easier.

Q: What kinds of image sequencing processes have you helped pioneer?
A: Starting as a graduate student at MIT in 1958, I worked on compression. So I developed a number of compression techniques for documents and images. Many of the ideas are used in international compression standards today. Another thing I did later was try to derive three-dimensional information from two-dimensional sequences, like finding the 3D movement from 2D sequences. Usually you don’t see the impact until many years later, so I think we have seen the impact on compression, but the impact from the work on relating 3D to 2D probably will become more evident in the next few years, in terms of applications to surveillance for example.

Q: You’ve traveled all over the world. Where have you gone? Why is this important to you?
A: I guess it’s a mixture of reasons. One is just my curiosity. But I guess I have been traveling mainly, well almost always, in connection with my work. I’m not just “taking vacation.” So I guess the reason is that research is a global thing. There are smart people everywhere, so in order to really keep a grasp of what’s going on and to get new ideas you need to talk to people in many different countries. But mainly, for me, Europe and Asia because they are more advanced in that area. Usually I go to a place only if there’s a conference or there is some university research lab.

Q: Tell me about a research accomplishment you’re proud of. How did this work impact society?
A: This is a difficult question because it’s like asking, “Which of your children do you love most?” It’s hard to say because there are different criteria to measure how important results are. One measure could be just the elegance of the theory. So many years ago I have proved some interesting theorems. It’s very elegant, but I think they’re useless. But I’m still proud of them. In terms of the impact, I think the two things I mentioned earlier—compression, where we’ve already seen the impact, and relating 3D to 2D. We have more and more information, and compression is needed for storing and transmitting it. Without compression, it’s really impossible. For relating 3D to 2D, I think the impact will come and interpreting what is happening in the 3D world based on 2D observations. There are many different applications for it.

Q: What do you enjoy most about teaching?
A: Actually teaching and research are really closely related. In teaching, of course I enjoy interacting with students. But also, when you try to teach something, you think about the topic really critically, and in many cases you’ll find something new. So teaching would lead to research, and of course, research will lead to teaching.

Q: What are you focused on today?
A: One of the most recent works we are doing is trying to implement a text messaging system, where the text will be delivered by a talking face. We synthesize the voice, and the challenge is trying to make it emotive— that is with some facial expression and a tone of voice appropriate to the content of the message. It turned out to be very interesting but very difficult. Almost all the synthetic speech systems commercially available are just neutral. So making it emotive is very difficult. It will be used in emails. So if you send an email to your friend, then on the other side it would be delivered by this. The kind of things we do are in two related areas. One is human-computer interface, so this is one example. The other area, which is also very important, is searching for information in a huge database, especially multimedia.

Q: What technology that’s currently under development are you most anxious/excited to see completed?

Again, it’s those two areas. One is the human-computer interface, and we would like to eventually have computers or systems that interact with humans in a more natural matter. The other is how to deal with the huge amount of information. One specific question is whether we can have a search engine that is more multimedia. Right now, you search by keywords on the Web. But there are all kinds of information in different forms, like images and audio. And how can we have a smarter search engine where you can search not just based on keywords, but also based on images by saying “I want similar images”. But one of the biggest problems to which I hope there is a solution is how we deal with this information explosion. We have more and more and more information, and we get junk email all the time. So I think that if we cannot solve it, then I think the world will collapse. We are completely buried.

Share this story

This story was published July 1, 2008.