DIY Book Scanner

Hi all,

Just trying to get Scan Tailor to include page numbers in the content area. Does anyone have any idea how to coax ST to do this, without having to manually drag the content area box boundaries up for each page (in a 600 page book)?

Thanks, I'm attaching a sample image.

First things first...are you using the latest version? There have been several updates that helped deal with this problem. There have been a number of threads about it... here's one where Tulon gives a hint: http://diybookscanner.org/forum/viewtopic.php?f=8&t=373

I'm not saying it's been solved, but it has come up before.

Also, I think cropping the wood and screws out of the picture will work wonders as well. Give it a shot, it can't hurt.

I tested the scan (p.11) you posted and the pagenumber was included in the selected area after I had converted the scan to greyscale. I noticed that the left and right margin of p.11 (i.e. top and bottom of the page) were significantly less well lit than the center (or is it the fault of the camera/lens, i.e. vignetting?). If the lighting is uneven, correcting that might improve your results.

You have a very low DPI there - maybe 70 or 80. That's way below what Scan Tailor will even accept. To make it accept those shots, you obviously had to specify a larger DPI than you really had, which makes content selection even less accurate.
The reason ST doesn't accept anything below 150 DPI is that you just won't get good results from that kind of input.
The way to go is to zoom in, if possible. I mean, your camera obviously doesn't have too many megapixels, so you don't want to waste any of them.

Hi all, thanks for the help.

Will try to zoom a bit more, and perhaps treat myself to better cameras in the near future (I'm using Canon A590s...).

RE: pixels, 1 inch of the book (about 3.5 lines of text because it is an odd printing) was about 158 pixels... Will have to go back and check again.

I also consistently have these problems, also at higher dpi's. I am just wondering whether there is anything useful one can do to preprocess the images for example in Lightroom to make it easier for Scantailor to detect the edges. As I don't know how the algorithm works, playing around without this knowledge probably won't help much.

In some cases you can do something, in others you can't. I can't tell without looking at the original shot. Lowering input DPI below the actual value may help sometimes.