The Big Squeeze: HARC's Video Compression Technology

By Brian Tyson

Sequence: Volume 30, Number 5

Release Date: September/October 1995

A new technology on the horizon could make universities resemble one-room
school houses . . . but that room could be your living room.

The concept of the virtual university where students learn by telecommuting
and videoconferencing is gaining momentum, but the technology is still
jittery. Setting up a classroom as a pseudo-television studio is expensive
and offers little communication between instructor and student. With limited
feedback, the instructor is deprived of the usual visual clues, such as
gestures and facial expressions, to determine whether a lecture on
quantum-mechanical tunneling is hitting its mark or veering into quantum
ho-hum land. The end product often becomes nothing more than a telephone
conference call with jerky screen resolution.

Catching a Wavelet

But this all may soon change. Dr. Andrew J. Blanchard heads up a team at the
Houston Advanced Research Center (HARC), a nonprofit university-linked
research institution, that has developed a wavelet-based compression
software that challenges conventional thinking about multimedia and could
easily make the virtual university a reality. HARC-C (for HARC-Compression)
software squeezes the huge amounts of data that make up a video file so
that, for instance, VHS-quality video can be sent to a laptop computer over
an ISDN phone line with full-screen, full-motion color at 30 frames per
second. The image is compressed by as much as 480-to-1 with only minor
deterioration. Under optimal conditions, HARC-C can handle NTSC 640 x 480
compressed video, full-screen playback at 38 frames per second in simulated
24-bit color. It can play direct from a hard disk or from CD-ROM with no
special memory requirements (it requires only 4MB of memory on PC).

What makes this compression technology impressive is that no specialized
hardware is required, it improves the speed of compression, and the results
are virtually identical to the original. Blanchard says traditional
compression technologies such as JPEG and MPEG can't come close to matching
his team's offering. "We're taking high-bandwidth information and putting it
through low-bandwidth lines," Blanchard says. "It can happen in real-time in
software, which is rare. Our technology has no hardware requirements."

HARC-C solves the problem of file size and limited network resources by
coding an image in a form that is easily stored and shipped. Building on a
set of mathematical principles called "wavelet" theory, the attributes of
each pixel are described mathematically in terms of frequency, energy and
timing. This approach enables the data to be compressed as a whole rather
than in discrete blocks, managing the task in real-time, which is about five
times faster than current technologies. Ultimately, the software could allow
Internet users to ship high-quality video as easily as they currently ship

Just Add Water

Texas A&M University professors Charles K. Chui and P.K. Yuen are the
originators of the algorithm that became HARC-C. Chui explains HARC-C in
terms of an orange. "Think of information like orange juice. If you want
orange juice for breakfast, but don't live where oranges grow, you could
import the fruit and squeeze your own juice. It is easier, however, to have
the oranges processed into concentrate, import that and add your own water
to make juice. You get the orange juice you wanted but don't have to ship
all the unnecessary parts of the orange."

It seems a shame that Chui's contributions to the field of mathematics are
reduced to the phrase "just add water." Yet, as complicated as the
mathematics are, the final product reduces storage, can be compressed at
different levels, and can be transmitted more freely�which uncomplicates
things quickly. And while a number of image compression technologies now on
the market require special computer hardware, HARC-C can easily run on a
variety of hardware platforms.

Teaching is a High-Bandwidth Process

The software holds a tremendous potential for higher education and the way
students can interact with each other and their instructors. Sitting at
home, a student with a one-inch camera mounted on his personal computer
could interact verbally and visually with the professor as well as the other
students�each of whom is also sitting at home in front of his or her PC. The
hardware�cameras, high-resolution video screens and computers�is available
for such a scenario. However, transmitting high-bandwidth information
through a low-bandwidth line poses the real dilemma, particularly when
graphics and video are involved.

Blanchard describes how an art instructor, for example, could take apart a
painting layer by layer to examine how the artist applied certain colors, or
used certain strokes. Views of the painting could be examined from different
angles. In film and motion picture studies, lighting and shadows could be
manipulated to see how the Hitchcocks and Truffauts crafted classic scenes.
"There are opportunities and concepts that will enable us to capture things
we never thought of," he says. The most immediate benefits will enable
students to conduct collaborative work, sharing a common window on a
workstation and both contributing.

University computer laboratories, most of which are wired by Ethernet or
better, need no alteration to meet compression requirements or
specifications to receive movies or video transmitted from one workstation
to another.

And what would artist Michael William Harnett think of what is being done
with his paintings in demonstrating the effectiveness of HARC-C? "I think
that if Mr. Harnett could get his painting disseminated to as many people as
possible he would be pleased . . . because most people can't visit the
Dallas Museum of Art," Blanchard says.

"Teaching is a high-bandwidth process. Just one 50-minute lecture has an
incredible amount of information. The problem is fitting that information
into a lower bandwidth frequency. Educators need to have low-bandwidth
connections so everyone can use it."

It appears that society is reaching a point where learning will be a
continuous process and can be acquired anytime, in any time zone, across any
geographical or political boundary. Blanchard says the real hurdle will be
cultural, not technological. "The reality is the cultural impasse. The
technology will be here long before the culture. But, like anything, the
economics of the situation will drive the situation. When financial
pressures force universities to compete differently in order to survive,
then the cultural barriers will come down."

HARC-C has been licensed by Ball Corporation's Aerospace and Communications
group to compress images generated by satellites before they are sent back
to Earth. HARC currently is negotiating with other companies for licensing
in other application areas. The software is anticipated to be available to
the education market in 10 to 12 months.

Brian Tyson is editor of the Corollary, a science and technology magazine
published by the Houston Advanced Research Center.

� 1995 Educom.

Take me to the index