Ishikawa Komuro Laboratory (the same lab that previously brought us the baseball pitching and batting robots) is now showing off another novel use for their high speed computer vision processing technology: book scanning. A camera sensor running at 1000fps captures the pages as they turn, recording the text and images despite any distortion and lighting differences. The system could be used to speed up the digitization process of low-cost e-books and other library data.
The camera uses lights connected to a synchronized control circuit and a laser range projector to estimate the three-dimensional page geometry. This allows it to correct any distortion from the page being turned while at the same time flashing it with uniform, ideal lighting. The 3D data can even be reproduced on a computer. The system could theoretically be used for color copying as well, but the current quality of the scanning data still requires some improvement. Google has been digitizing books and a system has been developed to automate page-turning to reduce the human labor component in this time-consuming, expensive line of work.
Perhaps Short Circuit’s Johnny 5, flipping through the pages of books, scanning them in mere seconds all while shouting, “More input!” isn’t so far-fetched after all…?