SOFTWARE ENABLES REAL-TIME VIDEO ON THE INTERNET
COLUMBUS, Ohio -- Ohio State University
researchers have developed software that transmits video signals in real
time over the Internet.
The software seems to provide a smoother and
clearer picture than other real-time video programs available. It
may one day enable practical video conferencing on standard desktop computers.
Because of heavy traffic, the Internet can’t
carry the massive amounts of data necessary for smooth, sharp video signals.
Today's software compresses video data into tiny packages and sends it
over the Internet, but viewers can only play it back after they've downloaded
an entire video. Not only is such video communication not real-time,
but the image lacks clarity. Ohio State researchers may have found
a way around this problem.
"Our goal was to compress the data and still
have very high image quality once the signal is reconstructed -- and to
do this
very fast, so that people on their computers can see a video of something
as it is happening," said Yuan F. Zheng, chair and professor of electrical
engineering.
For this work, Zheng and his colleagues --
students Cheng-Jiang Lin and Daji Qiao, and Chun-Li Pei, a visiting professor
from China -- took a complicated computer algorithm called a wavelet transform
and simplified it. Wavelet algorithms produce high-quality video
images, but take a long time to calculate on a normal desktop computer.
But Ohio State's wavelet algorithm computes much quicker -- 0.05 seconds
for a 2 inch by 2 inch video image, Zheng said.
The researchers are now making efforts to
identify the parts of a video image that change from frame to frame so
that the data can be further compressed. For instance, in a video
of a person speaking, only the person moves, and the background stays the
same. To save time, wavelet algorithms transform only the part of
the picture that has changed, and transmit only the data for that part.
While some computer chip manufacturers have
already developed chips specifically for wavelet compression, each chip
costs abut $100, and a number of chips are needed for a compression system.
"We feel that our software offers an advantage
because any desktop PC can use it without additional hardware," said Zheng.
"Basically, we modified a wavelet algorithm
so the computation is very fast for images measuring about 2 inches by
2 inches," said Zheng. "And we only need 0.05 seconds to compress
the data of a 2 inch by 2 inch window. That lets us show 20 frames
per second. We've overcome computation speed, so now we can use wavelets
do real-time video compression."
Compression makes the amount of data very
small. For example, the Ohio State software could compress data sent
over the Internet into packages one-tenth of their original size, so in
transit the packages would require only one-tenth of their original space.
Then a destination computer could decompress the signal and reconstruct
the image.
Since millions of people use the Internet
at the same time, phone lines become crowded with data and there's less
room for video to squeeze through. But because this software compresses
video data into a fraction of its normal size, Internet congestion becomes
less of a problem.
"Even with congestion on the Internet, we
can send more frames per second than uncompressed video," said Zheng.
And the more frames the software sends per second, the smoother and better
quality the image
The current prototype version of the software
can transmit a video image of about 4 square inches. In early April,
Zheng and his colleagues ran a successful demonstration of the prototype
software with another office across the university campus.
"We hope that in the future our software will
be as useful as live TV, but on the Internet," said Zheng.
Contact: Yuan F. Zheng, (614) 292-2571; Zheng.5@osu.edu
Written by Pam Frost, (614) 292-9475; Frost.18@osu.edu
Return
to the current month abstract page
Return
to the Research page
Return to the OSU Homepage
Go
to the Reasearch, Newsfeature, and Cancer Report Archive