A Leap for High-Speed Transmission

As the Web matures, the demand grows ever stronger to transmit more information in less time and at a higher resolution without creating cyber-jams of monumental proportion.

In an effort to meet that challenge, computer scientists at Caltech and Lucent Technologies Inc.’s Bell Labs have teamed up to create the technology to transmit even high-resolution images of three-dimensional objects at lightning speed.

The key is not faster modems or broader bandwidth, but the way the data are acquired, encoded and compressed, according to Peter Schroder of Caltech and Wim Sweldens of Bell Labs, leaders of the project.

Their research could soon lead to “virtual fitting rooms� where e-commerce customers can try on clothes and see how they look from every angle while sitting in front of their own conventional desktops.

That’s only one of many applications for the new technology known as digital geometry compression. Manufacturing, entertainment, education, medicine and other industries are expected to adopt the technology in the months ahead because it paves the way for three-dimensional imaging that will allow inspection of an object from every angle without clogging the transmission lines and filling up memory banks.

And here’s the neat point: The primary cost is at the supplier’s end, not the client’s. It won’t be necessary to replace the old desktop to take part in this major leap forward. But your local shopping mall or a major manufacturer will have to invest in some fancy 3-D scanners.

Transmitting large data sets always requires some compromise. High resolution requires huge chunks of memory, and transmission can be tediously slow, so some trade-offs must be made.

Using a standard 28.8-kilobit modem, for example, it may take less than a couple of seconds to download a page of text, but if the user wants a full-motion video, it can literally take days. To reduce the time required for transmission, computer scientists have developed several ways to compress files, generally by throwing out extraneous data, but that usually leads to a loss in quality.

Schroder and Sweldens are riding the wave of the future, or more specifically, the “wavelets.� This mathematical technique is rapidly replacing the current method of compression, known as Fourier transforms, because it leaps over a number of high hurdles.

Wavelets can be used to prioritize data, sending the most important stuff first and allowing the user to shut off the stream of information whenever he or she is satisfied with the resolution of the image on the screen. If you want to see how the new duds look, you can shut it off after a few seconds, but if you need a high-resolution image of an aircraft part, you can leave it on longer.

“At any given point in the transmission you can just hit the stop button, and then you can work with whatever you have at that point,� Sweldens said.

But equally important, wavelets allow the supplier to layer the data, building up a 3-D image with each new layer.

Sweldens developed that technology about four years ago, and he calls it “lifting.� It improves the image incrementally, thus leading to precise representations of 3-D objects.

“Originally, wavelets needed a fair amount of heavy mathematical machinery to understand how they needed to be constructed,� Sweldens said. “That scared a lot of people off.�

Lifting substitutes simple geometry for that “heavy mathematical machinery,� he said.

“If you are outside a building and you want to go up a floor, you take a ladder and you go up one step at a time,� he said. “Each step is very easy to build and implement. If you just cascade enough of these little steps together, you could get anywhere you want,� including a detailed image of a 3-D object.

Lifting also reduces the amount of data that has to be stored.

The standard for high-resolution images is MPEG4 technology, and the scientists say their “lifting� technology is 12 times more efficient.

“That means we can crunch this piece of data into a file that’s 12 times smaller and hence will download 12 times faster over the Internet than what the standardized method does today,� Sweldens said.

The mall of the future, and that may not be very far off, will probably have a virtual fitting room, the scientists say, where customers can have their bodies scanned with 3-D laser scanners that are already available from several manufacturers.

The customer will receive a floppy disk with all the necessary data, which could then be fed through a home computer to a store or manufacturer.

“The entire shape and size of your body� would be on the disk, allowing the buyer to see just which curves the clothes project, or conceal, before making the purchase, Sweldens said.

At this point, that disk would have to contain about 10 megabytes, and that’s more than a floppy could hold, but the scientists say their technology could get that down to about the size of a low-resolution JPEG file. JPEG is a type of image-compressing technique.

A manufacturer of automotive parts, however, might need to show higher resolution so that customers could put the parts together in a virtual setting to see which ones worked best.

Manufacturing companies that can justify the sizable investment in systems for 3-D scanning and digital geometry processing have already begun using the technology to create virtual parts catalogs, according to Caltech’s Schroder.

“They can use geometric representations when they put out requests for parts, use geometry to guide fabrication equipment and compare scans of newly made parts to the original designs,� Schroder said.

The cost of the equipment is a limiting factor for some manufacturers at this point, the scientists concede, but as the technology becomes more widespread, that cost should come down.

For most consumers, the technology should allow them to hold that priceless antique in their “virtual hands� and even see how it looks in their “virtual living room� before buying it, Sweldens said.

*

Lee Dye can be reached at [email protected].