Convention in Rochester draws top names in the field

SEPTEMBER 25--About 1000 of the world's top imaging scientists gathered in Rochester, NY, for the ninth annual International Conference on Image Processing, the largest and most prestigious meeting of its kind. The scientists came from all over the world to trade ideas on the latest and most-promising lines of research in still and video photography, aerial, space and medical imaging, and more.

Attendees Monday heard updates on two of the hottest areas of imaging research: the display of still and video pictures on mobile devices and imaging software that can identify an individual's face or fingerprint for security and other purposes. In mobile imaging, scientists are working to make images look better on small display screens in devices such as cell phones or personal digital assistants, according to Jiebo Luo, senior principal scientist at Kodak and a conference organizer.

Scientists are also working hard to improve paths of communication between the devices and computer servers via wireless networks, Luo said. With better communication, images will appear on the screens faster and more reliably with richer colors. "What we're trying to do is make sure the image is of the best quality, whether it's for sharing or to preview for printing," Luo said.

There are other technical hurdles that researchers are working to overcome, said Chang Wen, head of interactive media at Sarnoff Corp. (Princeton, NJ), which markets products for electronic, biomedical, and information technology fields. For instance, time is an even more precious commodity in delivering images and data to cell phones, Chen said. Internet users don't mind waiting a few seconds, but "with wireless, you want it to run right now."

The other hot area at the conference was software technology to identify individuals in images, said Tsuhan Chen, professor of electrical and computer engineering at Carnegie Mellon University (Pittsburgh, PA). That software is part of a broad field known as biometrics, which uses computers to recognize fingerprints, faces, voices, or other personal attributes.