Hello,
After scanning some pages I would like to improve the quality. When I scan a page often I can see the text on the back of the sheet.
After searching, it seems the problem has to do with the thickness of a sheet. To improve it, it's possible to put a thick sheet of paper or something
like that on the back of the sheet we want to scan. Then the text on the back will disappear. This method works for a flatbed scanner. But most
of the time I use an adf to scan. I have too many sheets to scan. I don't want to scan them with the flatbed scanner. Does it exists something to
hide the text which is visible from the back of the sheet when I use an adf ?
It helps to do some post-processing, but I also have some pictures on my scan. And in those pictures we see the text. On normal pages black and white
this problem doesn't exists. It's for my colored scans. Perhaps I can also buy something. It would be a possibility to use the adf to scan 10 pages and only the front and after that the back of those 10 pages. And each time I would add a layer to thicken the page. I'm open to all possibilities.
Thank you!
Back of page visible on the scan
Moderator: peterZ
Re: Back of page visible on the scan
Have you tried experimenting with the scanner settings to see if you can reduce bleedthrough from the back of the page without noticeably degrading the colour images? It is quite common for people to scan using the scanner default settings, without exploring the often quite powerful scanning options available in the scanner interface.
Alternatively, do you have a sample image you can upload to see if it can be enhanced using image editing software?
Alternatively, do you have a sample image you can upload to see if it can be enhanced using image editing software?
Re: Back of page visible on the scan
The trick is to put a black piece of paper behind the page, but with using an Automatic Document Feeder this wouldn't be worth the trouble. And quite frankly, it's not really worth the trouble to do it on a dual platen based book scanner either because you'd continually be inserting and removing black paper under the left and right pages prior to the shot.
For that reason I thought about doing this in a post processing phase using the image of the scan on the back of the page. There are some registration issues to deal with but it could work. I posted about this a while back in one of the threads on this website (http://www.diybookscanner.org/forum/vie ... 7&start=20).
For that reason I thought about doing this in a post processing phase using the image of the scan on the back of the page. There are some registration issues to deal with but it could work. I posted about this a while back in one of the threads on this website (http://www.diybookscanner.org/forum/vie ... 7&start=20).
Re: Back of page visible on the scan
Thank you for your answers. I have already thought to take a look at the settings of my scanner. I take a look again and I have found nothing to help me. But I think it would be possible to enhance the image with an editing software. My goal is to automate the process at the end, so the best would be a command line tool. I'm on linux, probably imagemagick is a good candidate.
For the scans: I'm not sure it's a good idea to post them here. Could you give me your email address with a private message and I'll send them by email ?
At the moment I think a problem is because I have increase the brightness during the scanning process. My goal was to get better colors and if I let the default settings, the colors don't render very well. But perhaps I could reduce the brightness and after the scanning process I could increase the brightness again and I would get better results. I discovered a few weeks ago the following command:
It's from imagemagick. With it we can improve the colors. If I reduced the brightness during the scan and apply the command after, I get a better result. I could reduce even more the brightness of the scan to reduced the bleedthrough even more. But at the moment I don't know how I can get good colors after that. Do you have any ideas ?
@dpc: If I understand your solution, it's for a scanner with a digital camera. If not, I don't understand very well. I'm using the scanner of my printer.
For the scans: I'm not sure it's a good idea to post them here. Could you give me your email address with a private message and I'll send them by email ?
At the moment I think a problem is because I have increase the brightness during the scanning process. My goal was to get better colors and if I let the default settings, the colors don't render very well. But perhaps I could reduce the brightness and after the scanning process I could increase the brightness again and I would get better results. I discovered a few weeks ago the following command:
Code: Select all
mogrify -normalize -level 10%,90% -sharpen 0x1 scan_1.tif
@dpc: If I understand your solution, it's for a scanner with a digital camera. If not, I don't understand very well. I'm using the scanner of my printer.
Re: Back of page visible on the scan
The technique that I'm describing would work with any scanning method, provided that you're scanning the front and back of the same page. Imagine a scanning a front page that is completely blank (white). And you scan a back page that has black text on a white page. In the post processing phase for the front page you look for black pixels on the back page and subtract the black from the front page at that point. The simple way to do this in to invert the colors of the back page then do an additive blend between the two pages.
As I mentioned before the registration between the two pages is tricky but you can perform an iterative process with lower resolution images to find the amount of shift between the front and back pages that results in the lowest overall noise (bleed-through). You also have to realize that if there is bleed-through coming to the front page from dark areas on the back page, you'll see the same thing from dark areas on the front page that bleed-through to the back page. You likely don't want to consider that darkened area when subtracting dark on the back page from the front page because black pixels on the front page should remain black.
As I mentioned before the registration between the two pages is tricky but you can perform an iterative process with lower resolution images to find the amount of shift between the front and back pages that results in the lowest overall noise (bleed-through). You also have to realize that if there is bleed-through coming to the front page from dark areas on the back page, you'll see the same thing from dark areas on the front page that bleed-through to the back page. You likely don't want to consider that darkened area when subtracting dark on the back page from the front page because black pixels on the front page should remain black.
Re: Back of page visible on the scan
Thank you, now I understand better what has to be done. I'll keep it as a possibility to reduce the bleedthrough.
cday has modified one of my scan and after changing the levels, he was able to eliminate the bleedthrough. So another solution
is to edit the image after the scan. A possible software is XnViewMP. With this software, the levels can be edited with the shortcut
'l'. There we can observe the change in real time. I'll experiment a little bit more and after that I'll have a look to see if a software
like imagemagick can be use to automate the process.
I have written all details here, perhaps it can help somebody else.
Thank you very much for your help! Problem solved!
cday has modified one of my scan and after changing the levels, he was able to eliminate the bleedthrough. So another solution
is to edit the image after the scan. A possible software is XnViewMP. With this software, the levels can be edited with the shortcut
'l'. There we can observe the change in real time. I'll experiment a little bit more and after that I'll have a look to see if a software
like imagemagick can be use to automate the process.
I have written all details here, perhaps it can help somebody else.
Thank you very much for your help! Problem solved!