The Hindu Business Line : The key: 10 lines of code

IF you graduated from an engineering college in 2001, there is a good chance that you are part of the huge workforce that keeps the wheels of the industry moving.

You are probably happy, on your way up the ladder in a Top 5 software services company. Happy... with the prospect that a project leader's position is within grasp.

Arvind Thiagarajan too, probably, started off with such thoughts when he got his bachelor's in engineering from Anna University that year.

But a simple, 10-line computer software program he wrote changed his life. He is now co-founder of a company that not only does business in parts of Asia, has not only attracted venture capital investment of $2 million from Korea, but is also listed on the Australian stock exchanges.

So what does his program do? It compresses images more efficiently than existing compression technologies do, says Arvind.

Why is this important? After all, it's easy to get bandwidth at lower costs now and downloading even movies on the Internet is not heavy-duty stuff anymore. And storage costs are also plummeting. Sure, if you thought that, that was intelligent thinking.

But then, however much storage costs plummet, some organisations need such huge amounts of storage space that any compression technology helps. And the more efficient, the better.

Think of hospitals. In several countries, the law requires them to store images of patient scans for a minimum of seven years! Think of scans, each of which measure 16 GB! (your average home computer comes with storage space of 80 GB.)

Storing 16 GB scans for numerous patients per day for a period of seven years is a stiff challenge. That was the challenge that a Singapore-based doctor asked Thiagarajan to surmount. And that was when Thiagarajan's tryst with compression started.

Says Thiagarajan, "10 lines of code was all it took. I helped the doctor reduce the image to less than 1 GB."

He says that his logic is more efficient than the JPEG 2000 technique of compression. Recent experiments that MatrixView, his company, has conducted show that for the average echocardiogram (ECG) image, his technique compresses the image thrice as much as JPEG does.

"Also, there is no loss of data in our compression technique." What this means is that in traditional technologies that aim at intense compression, there is always loss of some data.

When the image is decompressed for viewing again, the loss of data shows up.

Thiagarajan's algorithm helps you avoid that. The company, with a team of 50 researchers working out of India, is also working on audio and video compression methods.

For someone who so recently graduated out of college, isn't it a challenge attracting and retaining talent?

"Sure. Some folks might find it difficult. But that should soon change. Because of my work in the area of compression, I have been offered the opportunity to do a Ph.D program at a US university, with minimum class work. Once I get through with that, it should be easier to attract researchers with Ph.Ds behind them," he says.

In the world of business

Thiagarajan says that it was difficult to convince potential users. "There was disbelief that we could achieve such high conversion rates without data loss." Over time, that has changed, especially in parts of the Asia-Pacific outside India.

The company also tried to get funding. "The Singapore-based doctor put us onto a Korean gentleman who brought in $2 million in angel funding and then helped us go public in Australia." But it decided against true-blue venture capital funding before it went public in Australia. Says Thiagarajan, "We met several VCs. Many of them asked for an impossible stake in the company. They also laid down conditions that we look only at healthcare and no other space."

Thiagarajan and his brother, who runs the business side of things at MatrixView, decided not to opt for venture funding except for the individuals who had invested in the company.

As to going public, why Australia? "We found some customers there. With acceptance of our product increasing in that country, it made sense to list there."

MatrixView is now a Singapore-based company, listed in Australia, with a research team in India. The latest available figures are for the year ended June 2004, when it recorded revenues of S$ 28,530. After having tapped the Australian market for cash, it has cash and cash equivalents of S$ 3.7 million. It listed at A$ 0.50 and is now trading at A$ 1.50. Its market capitalisation has touched approximately $120 million.

The logic behind compression

Assume that you want to send in a series of numbers such as seen in the matrix, starting with the number 150 on the left top corner of the accompanying image. The numbers are read left to right and rows.

Here's how traditional compression has worked.

Repeat every number if it is not followed by an identical figure. Otherwise count the number of times it appears continuously and indicate as much.

In the above example, the compressed series for the first row would look like: 150, 3, 100, 3...

Thiagarajan's logic is different.

He says, use a matrix of just 1s and 0s to show the repetition (that he refers to as the bit plane) and another matrix to show numbers that do not repeat, as seen in the image.

Top picture by K. Ananthan

bharatk@thehindu.co.in

Article E-Mail :: Comment :: Syndication :: Printer Friendly Page

Stories in this Section
Let's negotiate

No let-up in sight

Hear me?

RDX — for mind-blowing cinema

The key: 10 lines of code

Making a difference

Removing malware

Installing Windows 98

Enabling anti-virus protection

`Drives will thrive'

Hold on to your laptop

All about spectrum

Monitor your customer's value

Building awareness