This person made a prototype of an image splitter software. With this you only need one camera.
https://www.youtube.com/watch?v=rjzxlA9RWio
He shared the code in the description.
I don't know how to use it though. Any help?
Image Splitter
Moderator: peterZ
-
granterpower
- Posts: 7
- Joined: 13 Dec 2015, 15:36
- Number of books owned: 0
- Country: Canada
Re: Image Splitter
I think there was talk about that tool in some older forum threads. He uses a Mac and you must install opencv to run it and then use something like Automator. I don't have a mac so can't say more about it. I guess you'd need a very good camera for the images to turn out ok since you capture two pages in one shot and then deskew.
It is worth mentioning that you can use only one camera even without that tool. You first do all the right side pages and then flip the book and do the left side pages.
It is worth mentioning that you can use only one camera even without that tool. You first do all the right side pages and then flip the book and do the left side pages.
Re: Image Splitter
I was just now going to make a new post about this, when I saw yours. I emailed the author earlier this week, asking him to put a license on the code (to make clear what he's ok with other people doing with it). He very quickly and kindly responded by putting an open and permissive license on it, meaning that others are free to develop with it as they like.
I forked the code yesterday, and have started cleaning it up and adding documentation to it: https://github.com/publicus/bookscan.
I can confirm that it works for me (on openSUSE Linux) with some test shots that I took. granterpower, what operating system are you running? Perhaps we could work together to figure it out for what you're running, and then add those instructions to the documentation I'm updating. What I'd really like is eventually to have users contribute build instructions for different systems, so that there could be (a) a central list of instructions for each platform, and, from that, (b) possibly a workflow to create ready-to-go compiled versions for the major operating systems (so that one could just download and run the program, rather than having to install OpenCV and the other dependencies).
I forked the code yesterday, and have started cleaning it up and adding documentation to it: https://github.com/publicus/bookscan.
I can confirm that it works for me (on openSUSE Linux) with some test shots that I took. granterpower, what operating system are you running? Perhaps we could work together to figure it out for what you're running, and then add those instructions to the documentation I'm updating. What I'd really like is eventually to have users contribute build instructions for different systems, so that there could be (a) a central list of instructions for each platform, and, from that, (b) possibly a workflow to create ready-to-go compiled versions for the major operating systems (so that one could just download and run the program, rather than having to install OpenCV and the other dependencies).
Re: Image Splitter
It's also worth noting that I took my test shots in low light, so they were blurry. Thus, the program didn't always recognize the glyphs that I'd put on the platen. Having said that, when it did recognize them, it worked really well -- I took a test shot at perhaps a 75 degree angle from a page, and the software de-keystoned it in less than a second such that it looked completely normal.