ACM TechNews

Photo Industry Braces for Another Revolution

CNet (03/03/08) Shankland, Stephen

The exploitation of how computers can use sensor data either within the camera or on a PC forms the basis of the next photo industry revolution, which researchers are calling computational photography. One notable research area is the use of a computer to stitch together multiple photos into a single composite image of the same scene. MotionDSP software combines multiple images by exploiting the fact that multiple frames of a video record the same subject matter, and processing that can generate an image whose fidelity is superior to that of any individual frame. MIT's Rob Fergus has been developing software to eliminate blur in photos distorted by camera shake, examine photos to deduce how the camera shook when the picture was taken, then facilitate those changes. Another area of research involves the employment of computational processing to render a scene in three dimensions, and potential applications of this technology include not only 3D hologram images, but making the camera capable of optimizing focus and exposure for each individual shot. Stanford University researchers have devised a camera that can gauge depth through the use of hundreds of minuscule lenses over the sensor pixels, while Refocus Imaging has created a technology that produces data files that can be processed to focus the camera after the shot has been taken. The foundation of Refocus Imaging's technology is the light field concept, and Adobe Systems' Kevin Connor thinks light field technology will be a built-in component of cameras. Adobe has developed a prototype camera with a plenoptic lens that can help produce a 3D representation of a scene via the processing of the subimages captured by the lens' constituent lenses.

http://www.news.com/8301-13580_3-9882019-39.html

---

© Copyright 2008 Information, Inc. This service may be reproduced for internal distribution.