PDA

View Full Version : Book Scanner



RobotWorkshop
12-13-2011, 07:20 PM
Just came across this and though it looked like a pretty cool project:

http://www.diybookscanner.org/

mindrobots
12-13-2011, 07:43 PM
This was an interesting presentation at the Open Hardware Summit (watched it over the 'net). The impact they had in Haiti in helping to scan in medical records was impressive...a 'feel good' moment!

There could be bunches of local "Project Gutenberg" efforts popping up with this.

Loopy Byteloose
12-14-2011, 05:09 AM
This is a very worthwhile DIY project that gets around some rather tedious barriers provided by an ordinary scanner.

softcon
12-14-2011, 03:11 PM
This is exactly why I decided to begin building my own hardware. Projects like this book scanner is great. There's commercial products folks can purchase if you're blind to do this sort of thing, (https://www.kurzweiledu.com/store.php kurtzweil reader being the most popular at nearly $1,000)
http://www.irislink.com/c2-1982-189/1011-RI-Cardiris-us.aspx?adwp=GGS-RI-US&gclid=CM_xqNr2ga0CFVGz7QoddCy4SA read iris pro weighing in at $125 for a slimmed down version, but $471 corporate version. Then there's http://www.hamrick.com/ vuscan for 40 bucks for the standard version, or 79 for the pro version. At least vuscan is cheap enough for the average person to purchase on their own, w/o assistanc from state organizations. Unfortunately, all of these options assume you already own a flatbed scanner to use with the products, there are stand-alone products such as reading edge and older versions of the kurtzweil reading machine, both of which are much much more expensive. (Most adaptive products are) so building my own things is a much better deal.
I'll definitely be watching this site with interest, and if I can have my propeller offload some of the processing, I'll certainly give it a try. :)
Thanks for bringing this one to my attention. Great stuff here.

David B
12-14-2011, 04:21 PM
I like this project. I've been thinking of building a book scanner.

I've been experimenting with the Parallax 128 cell array mounted on a homemade scanner. It does a pretty good job of scanning a single line of text at at time, and I've have started work building a second linear translation mechanism to be able to scan a full pair of book pages, but it sure would be great to be able to scan a full page at a time.

It looks like most of these DIY scanners hack a Canon camera to get it to save raw images. That's a nice way to get a relatively inexpensive, large (like 3k by 3k) 2D array of optical sensors along with lenses, focusing and sensor control,

As far as I can tell, it looks like this Canon hack still results in saving images on the SD card, which then must be manually copied to a PC for additional processing, OCR or whatever.

It sure would be nice to get access to the images as soon as they're written to the card so OCR could be done simultaneously with the scanning.

Maybe it could be dome by emulating an SD card with an external circuit with its own microcontroller and a few megs of RAM, and plug that emulator into the Canon!

Skoorb
12-14-2011, 06:09 PM
Non-destructive scanning, now that's pretty cool.

I knew a guy 10 years ago that would put a book in clamps, run the spine down a belt sander to remove it, then feed the book through a scanner. The process was fast as hell, but of course the book was essentially destroyed in the process.

Loopy Byteloose
12-15-2011, 12:59 PM
The reason the authors seem to prefer Canon cameras is for an audio/video interface into a PC. I guess this can download images directly onto your monitor for positioning, and then a capture image can be file transfered.

It seems with the traditional scanner, the USB port bottlenecks the whole process and there is rather a lot of waiting. If I can take 100 double images a minute, an average 300 page book is scanned in less than two minutes. If you have to transfer files to the PC with SDcards rather than by USB, you still gain tremendous productivity. Then there is clean up software that does something like ImageMagik, which can run whole batches of video formated files to clean up rather quickly.

The only drawback is that the scheme uses TWO cameras with a simultaneous trigger. The trigger will first set focus and then also take the picture.

I may just build a ONE camera bench to replace my current scanner which is slow, has a nasty proprietary software with it, and requires that the computer be actively occupied while I am scanning. Building a cage that supports a camera from above and pointed directly down at a flat well lit surface can to the trick. And this kind of a set up can be used for other things - such as macro-photography, photos of 3-D small objects, and even some traditional animation. It could be used with a sheet of glass in some cases as a less adequate book scanner - for just a few pages rather than the whole text.

David B
12-15-2011, 03:52 PM
Do you have any information on getting the images out? The only live camera data outputs I've read about are NTSC/PAL compatible, which don't have anywhere near the resolution needed to get a good scan of a full page of text. I assumed that these book scanners convert the Canon to saving raw pixel data rather than a compressed JPEG, but only are able to store the image on an SD card, but I don't know that much about the Canon hack.

I'm about to go on vacation with my daughters, and have just learned that my older daughter has a model of Canon that has a hacking kit developed for it! Heh heh heh! The modifications have assurances that they are completely nondestructable to the camera. so I think I'll see if I can "borrow" her camera, sneak in my hacked SD card and see if I can get it to save a raw image; learn a little about this hack.

It seems like most successful book scanners rely on allowing the book to remain in a V orientation rather than being forced flat. That means that the scanner needs to either have two separate cameras or one camera on a moveable mount.

I haven't seen any automatic page turners on these projects - don't most of them rely on a human to manually turn the page? That would reduce the scan rate to more like 5 or 10 seconds per left and right page pair, or about an hour to scan a book.

Loopy Byteloose
12-16-2011, 05:37 AM
There seems to be some obvious areas of omission in their presentation.

1. No real discussion of automated page turners - but it could be done if you really needed to.
This would also require the bed to realign to center automatically after each photo as the center moves laterally as you work through a book.

2. The V configuration (a true 90 degrees) is intended to keep the pages as flat as possible while providing the maximum surface for photo work.
I was a bit disappointed to see that they generally use two cameras and that some of the cameras are $300USD each.

3. Software is in the public domain and freely available. But they don't go into great depth about what it can do. Much of the software is also intended for stereo optic photography. So you could use the two cameras on a bar to create that technology as well. They did NOT mention a personal favorite of mine, called ImageMagik that will clean up and reform batches of photos or graphic images.

And so, if you really want to do this with all the bells and whistles - it will require some exploration, thought, and planning. Personally, I'd be happy just flipping pages and tweaking images as I go. The quality might be better and you still can do a 300 page text in a rather short period - maybe ten minutes.

I think that the default is too have all and everything in a JPEG format and to download SDcards directly to your computer for manipulation after you have done a photo run.

Rather than move the camera, you can run through the whole book for odd or even pages and then TURN the book around and do a second run. The set up could be built to eventually allow for a second camera, but work fine with just one. It would be rather silly to move the camera everything you turned a page.

It seems that the NTSC/PAL video output would be used ONLY to properly frame and focus the camera lens; not provide the final image.

Loopy Byteloose
12-22-2011, 04:55 PM
I have been thinking quite a bit about this DIY project and exploring what is the best way to go.

RAW file output is indeed available with the 'hacking kit' on an SDcard that boots into certain (not all) CANON digital cameras. It also seems that CANON digital cameras offer a remote shutter control, which is important (especially if you want two cameras to shoot the same picture at the same time). And CANON cameras offer a NTFS/PAL output to allow focus and alignment verification before shooting the photo.

All these features together make some of the CANON digital cameras wonderful for bench work and adapted to being a robotic hacker's accessory of choice.

The last thing that I was concerned about was camera mounts. IF I were to build such an exotic bench for camera work, the camera mounts would have to be solid and adaptable to whatever changes in book size occurred. I saw some photos that actually used a goose neck mount to hold the cameras and I gasped at such hardware. After all goose neck mounts are really easily bumped and likely to never get back so something near to an original placement.

The good news is that there is a huge variety of cheap camera mounts both for photography and CCTV these days. And if one were to consider building the frame out of pipe, there are even mounts intended for bicycles and motorcycle handlebars.

So I am at the point that I am shopping for the right CANON for about $100 USD locally. The camera mounts can be gotten easily via Amazon. The hack for the CANON firmware is on line and referred to in DIY book reader web page. I just have to verify that the camera has a remote trigger feature as well.

This seems to take a lot of shopping in order to really get something worthwhile, but since the camera can be also used for other projects (and two cameras can also be used for stereo photography) it all seems to be offering a very HIGH amount of fun for the technically inclined DIYer.

DavidSmith
12-22-2011, 10:39 PM
Skoorb,

Figure out a way to remove the pages and reassemble the book and EVERYONE (except the book binders) will love you.