Computers take up lip reading

Anyone remember the heart-chilling scene in 2001: A Space Odyssey when HAL (the intelligent computer) tells Dave that, despite all the precautions taken, he knows Dave was planning to disconnect him because he could read lips? Well today we’ve moved one step closer to that possibility.

New software code has been released that researchers claim allows developers to build computers that can see. Researchers say the code will lead to new developments in human and computer interaction.

Until today, computer vision applications have been restricted to two dimensions (2D) – so scenes are recognized as single, flat, colour fields. But stereoscopic-enabling source code, released by researchers at Intel, enables computers to recognize ‘depth’ as well as ‘flat’ images, therefore computers are able to break out individual objects and surfaces in a scene.

The new stereoscopic code is an open standard upon which commercial applications may be built. The code enhances vision applications such as gesture recognition, object tracking and face recognition.

In the future, those technologies may lead to more sophisticated computer interface methods, better security systems and biometric tools, improve robotics and space exploration methods, claim researchers.

Over the next five to 10 years, Intel Corp. expects computer vision to play a significant role in simplifying the interaction between users and computers.

“Intel wants to make computers aware of the visual world,” said Justin Rattner, Intel Fellow and director of Microprocessor Research, Intel Labs. “Until we introduced OpenCV last year, the lack of common tools kept this from happening.”

OpenCV, or Open Source Computer Vision Library, is a resource tool created by Intel, where the stereoscopic code is now available to researchers. The OpenCV library is intended to expedite efforts of researchers in industry and academia and lead to the development of commercial uses of computer vision.

Intel expects the code, and other resources on the library, to increase innovation and ultimately lead to the development of computer vision applications such as face recognition, gesture recognition, camera calibration, motion tracking and object identification technologies by PC manufacturers, camera vendors, and software companies.

Factors such as faster microprocessors, falling camera prices and increased video capture bandwidth from technologies like USB 2 are enabling real-time computer vision algorithms to run on standard PCs, Intel officials said.

The library is a toolbox of 500 imaging and computer vision functions and applications, available for Windows and Linux operating systems, and is free for commercial or academic use.