91av

Projecting a better image

Bob Johnstone discovers a research strength

EARLIER this month the director of Motorola’s research centre in Sydney
together with the head of one of its crack research teams made a whistle-stop
tour of Taipei, Hong Kong, Shanghai and Beijing, all in less than a week. Their
main mission was to find customers for a digital chip the centre has developed
for cameras.

The chip is one of a new breed of image sensors revolutionising the digital
camera market. Motorola’s new image sensor offers several important advantages
over the conventional electronic eye, the charge coupled device (CCD).

The chip consumes between one sixth and one eighth less power, for instance,
a key consideration for battery-powered devices like digital cameras. It is also
fabricated using the semiconductor industry’s standard process technology,
complementary metal-oxide-silicon (CMOS), which makes it easy to integrate with
other chips, such as memory chips. This in turn reduces the cost of building
camera systems.

But CMOS image sensors historically have suffered from several inherent
difficulties which reduce the sharpness of the picture they produce. To help
colleagues at the Motorola lab in Chicago solve these problems, the visual image
processing laboratory in Sydney developed a collection of algorithms that sit
between the input and output portions of the system. The result is a dramatic
improvement in image quality. “It’s only a matter of time now before CMOS
sensors will be as good as CCD sensors,” research team head Xing Zhang
predicts.

The product is a remarkable achievement for a country that makes neither
chips or cameras. Australians, however, do know a thing or two about the “black
maths” of digital signal processing. “We have world-class signal processing
engineers,” Zhang says, “and we are the best signal processing research team in
[the] Motorola [organisation].” These skills in image and video processing were
one of the main reasons the giant US firm set up a research centre in Sydney six
years ago.

Such expertise is not acquired by accident. Rather, it is the result of
intensive research activities over the past 10 to 15 years in fields like
satellite remote sensing, by organisations such as Telstra, the CSIRO, and the
Australian Defence Forces Academy. Zhang studied at the Academy in Canberra when
he first came to Australia as a post-doctoral fellow in 1988. He still maintains
connections with his former colleagues through regular visits to his native
Shenyang, China’s fourth largest city, in the far northeast of the country.

Zhang’s latest trip began in Taipei and Hong Kong, where Motorola has product
development groups. Then it was on to the mainland, to quiz China’s six digital
camera makers about their requirements for image sensors. “To guide our research
long-term,” research centre director, Peter Beadle says, “we need not only
clever engineers who can come up with ideas for the future, we need input on
what the market is actually doing.”

For example, they wanted answers to questions like, ‘Are cameras that take
both still pictures and short video sequences (such as the latest Japanese
models) going to become more dominant?’ If so, then “that has a very large
impact on the