Minimal equipment iPhone-based scanning
Posted: 21 Aug 2018, 11:05
Hi all,
I'm working on an iPhone-based scanning platform, mainly targeted at textbooks. The eventual idea is that you'll be able to scan books with just a phone and a car-style phone mount, with motion detection as you flip the pages. Of course, this means no platen, and so dewarping will have to be done in software somehow. For now, I've been working on a Python implementation of the text-line-based 2015 Kim/Koo/Cho paper which has the best current performance. I have that a bunch of other scan-related tools on my github, https://github.com/phulin/rebook. Theoretically, the dewarping should soon be much easier: the new iPhones (currently X/7+/8+) have a depth camera, which unfortunately I can't test as I don't have one.
My rudimentary app detects motion and takes pictures, but doesn't do any post-processing. Because there's no platen and no trigger, it can reach very high speeds - I've been getting ~2000 pph when taking photos of dual-page spreads.It saves the images to a zip file which you can then airdrop or otherwise transfer to a computer to run the python stuff. If you'd like to beta test the current app, send me a PM with your Apple account email. Sorry, no Android for now
This is all pretty rough - especially the python stuff - so you'll probably need some amount of expertise to work around the kinks.
Here are some example images:
Original: https://www.dropbox.com/s/fuux4mv8r11vy ... t.png?dl=0
Dewarped: https://www.dropbox.com/s/bcaf4a1k8qxkq ... 0.png?dl=0
I'm working on an iPhone-based scanning platform, mainly targeted at textbooks. The eventual idea is that you'll be able to scan books with just a phone and a car-style phone mount, with motion detection as you flip the pages. Of course, this means no platen, and so dewarping will have to be done in software somehow. For now, I've been working on a Python implementation of the text-line-based 2015 Kim/Koo/Cho paper which has the best current performance. I have that a bunch of other scan-related tools on my github, https://github.com/phulin/rebook. Theoretically, the dewarping should soon be much easier: the new iPhones (currently X/7+/8+) have a depth camera, which unfortunately I can't test as I don't have one.
My rudimentary app detects motion and takes pictures, but doesn't do any post-processing. Because there's no platen and no trigger, it can reach very high speeds - I've been getting ~2000 pph when taking photos of dual-page spreads.It saves the images to a zip file which you can then airdrop or otherwise transfer to a computer to run the python stuff. If you'd like to beta test the current app, send me a PM with your Apple account email. Sorry, no Android for now
This is all pretty rough - especially the python stuff - so you'll probably need some amount of expertise to work around the kinks.
Here are some example images:
Original: https://www.dropbox.com/s/fuux4mv8r11vy ... t.png?dl=0
Dewarped: https://www.dropbox.com/s/bcaf4a1k8qxkq ... 0.png?dl=0