Wikipedia says Google is using a special camera that’s able to scan books at a rate of 1,000 pages per hour, which doesn’t sound bad at all (I am talking about Google Books of course, the company’s online collection of digitized books ). But now a team at the elite University of Tokyo has announced the development of a device that can scan a 1,000-page book in four minutes.
The core component of the device is a high-speed camera that can make 500 shots in a single second and is based on infrared laser technology. Users are required to manually flip the pages under the camera, which then makes shots of the print material. The device is able to scan everything that’s printed on paper, from character written in latin or other languages to graphs or photos.
The researchers say they hope to sell the technology to libraries or publishers and rely on cooperations with copy machine makers to market the device.
Via Nikkei [registration required, paid subscription]
“Users are required to manually flip the pages under the camera, which then makes shots of the print material. ”
Show me someone that can flip through a book and align it for the camera at the rate of 500 pages a minute…
From the picture it looks like they fan through the pages. Which of course introduces no distortions in the images.
The biggest concern here is keeping the books intact. When you have a delicate volume you can’t flip the pages recklessly. That’s always been the trade off with high-speed book scanning — how to handle the books/pages while minimizing scan time.
Very clever OCR, I imagine. If it were intelligent enough, you wouldn’t have distortion in images. When the camera takes an image, maybe the software can determine the “edge” of paper and adjust for the parallax accordingly. And I also imagine it could read the page numbers as it scans to determine if any have been missed. Very cool, if so.
There’s only so much distortion you should try to accommodate — plus you want to provide the best image possible to any OCR algorithm. Scanning page numbers would be smart though.
“Users are required to manually flip the pages under the camera…” as bots still haven’t mastered the art of licking their thumbs.