Google has been scanning books into it's online databases for some time now. However, the method used to do this has been kept quiet. The University of Michigan has partnered with Google to scan the collections of printed material in all of the libraries on campus. Yet even I as a student who has gone on multiple library tours knows nothing more than Google has an automated system to scan the books.
Finally, a patent application from Google has revealed the book-scanner. A common issue with copies and electronic scans of books is that the section of the page near the spine is distorted. Google's system is designed to correct this. The scanner uses an infrared source to cover the book with infrared light. Two infrared cameras then form a 3-D model of the book. The cameras detect the shape and angle of the book to allow the optical character recognition (OCR) software to correct the distorted images. Adjusting the images for the distortion allows the OCR software to interpret the letters more accurately.