Flip-book Style Digitization

[youtube=http://www.youtube.com/watch?v=tCOXC5PTJj8]

This method of book digitization allows you to scan an entire book by fanning through the pages. It uses a high-speed camera that captures 500 frames per second to get a good look at each page. Processing software isolates each pages, analyzes any curve in the paper due to the flipping, and smooths out the image for better optical character recognition results. The greatly reduces the time it takes to digitize a book, even compared to setups that automatically flip pages.

[Thanks Erico]

15 thoughts on “Flip-book Style Digitization

  1. Nice, but I don’t think you’ll be able to use it on older books, the kind in archives, unless this thing can read dust. Could come handy in photocopiers though, get rid of the sliding bar.

    @Sam – That looks like it was shot in the Worlds Biggest Bookstore in Toronto, wish I could let this scanner loose in there for a few hours, think of all the money I’ed save!

  2. Well, guess they’ll have to write new legislation dictation who can and can’t have a book digitizer now. The money I could have made in college with this thing….

  3. I’m very impressed with the project. I wonder how it will work with non-straight text. There are plenty of magazines and layout books with a fluid layout that might trick the system into thinking the page is not flat, or a different shape.

    I almost think it should overlay a projected grid to help gather shape.

    Thanks for the video, it was a great watch.

  4. @mess_maker

    It doesn’t use the ‘straightness’ of the text to work out the curvature of the page. A series of horizontal lines are projected onto the page and an image captured, it is these lines that are then used to correct the curvature from the flipping.

  5. @AnthonyDi: the dude just built Johnny Frickin’ Five, and you’re worried about pages sticking together? Off the top of my head, I’d say OCR the corners looking for page numbers and make sure you got them all. Should be easy after going to the trouble of correcting for page bending, etc.

  6. This is a great idea, but the approx. 40-50 dpi (1000×1000 dpi camera taking in an approx. 2′ square FoV) seems a little low for OCR, especially when you factor in the curvature correction.

    My OCR software balks at 200 dpi…

  7. Screw OCR, I just want to dump my book pages into a PDF and keep them on my laptop!

    Textbooks are a racket IMO, especially now that a majority of the ones out now don’t come as hard back. so for the price I used to pay for a hard back book I have to purchase a paperback that will not hold up to abuse. plus some books you cannot even sell back because they have an on line supplement, and most book stores will not buy those back even if the card with the key has not even been opened yet!

    I spent 275 bucks on a soft back book that I only used for 1 semester and I could not sell it back to the bookstore because of the supplement that it came with even though we never used it in class

  8. I’m way ahead of you. I used to make copies of textbooks with a two camera setup and a remote control. 7-12 Megapixels is all you need, turn the contrast up a notch. They usually have copies of all of the required books in one of the campus libraries. Spend an hour scanning a science textbook, save $260 not paying the textbook cartel, (that’s a $520 loss if you include their gain and your loss), return the book to the library. 260 bucks an hour is way better than any campus job pays. You don’t need a ten thousand dollar high speed camera.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.