Pretty cool. I wonder if some pages don't show up clearly if you don't flip the page right. This has potential to save a lot of money for libraries and other learning institutions.
the xray i had in mind not only scans through the cover but through to each page it can xray scan not only 1 book but 10 at a time and if the machine were bigger there is no limit to how many books it can scan at a time if you build 1 big enough to fit a thousand books it would scan every single page all at once
@fiedel The way I understood it is the doughnut lights are simply to light the page, then on the alternate frames the laser shines the series of lines on the pages as seen at 0:10 As the computer knows these lines should be a certain length, straight and evenly spaced it can model the distorted shape of the page by how these lines are distorted, and hence correct the image. Thoughts?
It often happens that you skip a page by flipping with your thumb so when missing a page while using this would mean that you have to check if you havent missed any pages and if you did then go and take the pic of that again. So much for high speed scanning if you have to take 5x as much time to fix it.
I dont know anything more about scanning beside basic stuff, and this seems like nice idea to be worked on. AN IDEA! There is a long process from it to complete product. Often the most simple ideas can be developed into something great. At least thats how Sony made walkman. Its a schoolbook example of awesome device rated as uninteresting by test subjects in early phases. And Morita only wanted to listen some music while he was jogging or whatever :)
Impressive work. I did not fully understand the shape-recognition part of the show. Wouldn't the ringshaped flashlights have to project some sort of pattern? But Yoshihiro san says it's "just normal lightsource". I guess thats one main ingredient of the secret sauce ... or just lost in translation?
@SakuraHirot87 I understand what you are referring to, but if you do that many times, it's unlikely that the same page will be skipped each time. So imagine flipping a book ~10+ times and uploading all of that into the computer program which compares each data entry and compiles it into one file (examining number of pages, and adding in missed ones, etc). So the likelihood that the pages that were stuck together would be each time is very unlikely.
The trouble is, you can see that, as he was flipping through, a number of pages simply weren't showing the full text to the camera. No matter how fast or how sophisticated, if the computer isn't seeing the complete text, the scan's going to be incomplete. A person reading a sentence that has missing or partial words might be able to infer the meaning through the context and fill in the missing words. Computers that don't register the meanings of the words can do that.
This is cool guys! Hey keep your stereotypes to your self. These guys are actually making things, unlike us living in the USA sitting on our ass eating donuts.
This would make Google's digitization efforts ALOT easier.... I always thought of this, but i never thought it would be feasible. well a high speed camera and some great software have proved me wrong.
Awesome! Even if there isn't much application in scanning, just think of what the recognition software itself could do. I can see this having applications in artificial intelligence.
Uh. It's the University of Tokyo. That's in Japan, where they speak Japanese. His English is actually pretty good for someone who lives in Japan. I used to tutor English for ESL students.
@thisisjustjeff yes this is true but the fact that a college student made it and put his time and effort into it so dont bash it. Most college students cant even do a simple power point with out procrastinating.
This is awesome because the hardware isn't heaps specialised. I look forward to someone puting some GNU software out there that will do this. and an arduino controller perhaps?
Amazing technology guys, just a little advice, make a more dynamic presentation to make your video more attractive, I wouldn't have watched it if I had not seen it on an article :) 5 stars anyway
They actually hold a ton of US Treasury dollars (as every other country) because it has a high stable intrest rate. Also, the real intrest rate in Banks in Japan is 0%, so there is no reason to keep money in a bank there.
1:24 and you see it rendered a full 3-d image from a photo, so this technology can be using to help 3-d modelers. In addition, this can help 2d animators in doing a pencil test, this will speed up the process of traditional or tradigital (traditional and digital hybrid) animation. The example in vid is not that great; a book is not the greatest use for this, however, if you think of bigger possibilities then you may understand the true potential of this technology.
The image noise level must be high for this to work...an ISO of at least 1600...meaning the scan images will be grainy and adjusted by auto high level of softening light spectrums. My guess is the book scans will be good for light work and not for proper archiving because the eyes will strain too much.
Yea, it turns out English is one of the ONLY languages that has a "TH" sound. (the use of the tounge in Japanese is EXTREMLY limited) At the same time, I would like to see you make the Sound "TSU" and be able to use it in words. It is one of the hardest sounds for English natives to say.
Great work over in Tokyo as usual. Now sell it to Fujitsu so they can commercialize it. You'll be rich, I'll have one, and I can give my books away (while keeping digital copies). Everyone wins!!
This is wonderful. I wonder if this guy is familiar with Ben Goertzel and Peter Voss's AGI project? If not it's a marriage made in heaven = a book reading AGI.
@thisisjustjeff Yes, but do you realize how much FASTER this technique will be in comparison? It will really allow digitalizing all scriptures of humankind way quicker by simply flipping the pages like that rather than scanning the pages one by one.
the only problem i see with this is that when people flip pages they tend to mess up sometimes and clump pages together, there should be like a mechanical flipper or something to go with it
@bernie23232323232323 its not taking picture, its scanning, yeah u can do it with your own digital cam, but takes forever with that you just flip the pages takes seconds to scan full book
@mastershake1000 Just get a doublesided documentfeed scanner like the Canon DR-2010M, it scans 40 pages per minute. You have to destroy the book to get the single pages of course...
I'd imagine you'd flip through the book a few times thus eliminating the accidental skipped page. Unless, for some reason that page was skipped every time... and then I'd imagine no one would want to see that particular magazine of yours anyways. =)
look everyone i dont know what the machine is called all i know is you put up to ten hard back covers in and press the button and in 1 second you have captured all 10 books you then take out your usb stick and plug it inyour pc and you can access it all and it is clear as this text every single page !!