I love simple and clever solutions to seemingly difficult problems. Like how to scan a book and transform the image to digital text..
Traditional scanners require flat pages pressed to a transparant plate, but that is impossible with a book without damaging it. It is also slow and time consuming to press pages against a plate, scan them, remove the book from the plate, flip a page and repeat the process.
If you open and handle the book and the pages as if you are reading it yourself it is easy and fast. You can take a picture with a camera of a page, but the image is severely distorted so the error-rate of OCR programs skyrockets.
Google has come up with the solution. Ingenious: Occam’s Razor applied to an engineering problem.
An extra infrared camera-set creates a 3-D image of the curved bookpages. This information is used to correct the image of the bookpages for spatial distortions. And voila, the OCR programs can work like advertised.
