First build with automatic dewarping

Scan Tailor specific announcements, releases, workflows, tips, etc. NO FEATURE REQUESTS IN THIS FORUM, please.

Moderator: peterZ

Post Reply
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: First build with automatic dewarping

Post by Tulon »

The first one is probably hopeless, matt. There are no long lines at all. Those separator lines would do, if they didn't have text so close to them. The second one might be fine, after I finish a better selection method for representative lines.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
matt

Re: First build with automatic dewarping

Post by matt »

Thanks for the input. Also, I'm curious about using the top/bottom edges of the page (if available) as cues for dewarping. I know that another image processing program (Snapter) does this; in fact it requires that the top/bottom edges be visible. Would that sort of grid algorithm be helpful/possible in the first example page I posted?
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: First build with automatic dewarping

Post by Tulon »

matt wrote:in fact it requires that the top/bottom edges be visible. Would that sort of grid algorithm be helpful/possible in the first example page I posted?
I don't want to require that, and neither do I want to invest another few months to come up with an edge tracing algorithm.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
eL_PuSHeR
Posts: 125
Joined: 28 Jun 2010, 15:25

Re: First build with automatic dewarping

Post by eL_PuSHeR »

Report: 1.0 ßeta5 doesn't hang on me anymore.

Great work, Tulon.

Best regards.
matt

Re: First build with automatic dewarping

Post by matt »

I don't want to require that, and neither do I want to invest another few months to come up with an edge tracing algorithm.
Gotcha. In my naiveté I thought edge tracing might be a special case of some more advanced algorithm that was already in use.

On another front, I"m noticing that ST (latest from Git) is crashing during batch output processing with auto-dewarping enabled. When this happens the following is displayed at the command line
Qt has caught an exception thrown from an event handler. Throwing
exceptions from an event handler is not supported in Qt. You must
reimplement QApplication::notify() and catch all exceptions there.

Abort trap
and a standard Mac crash reporter pops up and offers to send the report to Apple. Will you get this information as well? If not, would you like me to private message or otherwise send the report to you?

Thanks!
Matt
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: First build with automatic dewarping

Post by Tulon »

matt wrote:and a standard Mac crash reporter pops up and offers to send the report to Apple. Will you get this information as well?
Nope.
matt wrote: If not, would you like me to private message or otherwise send the report to you?
Yes, PM is the way to go.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: First build with automatic dewarping

Post by Tulon »

Beta6 has arrived: http://depositfiles.com/files/y5en4ust4

I implemented RANSAC for selecting representative lines and made some other improvements. Here are a couple examples where beta6 succeeds where previous versions failed:
good1_lines.jpg
(80.08 KiB) Downloaded 5385 times
good1_result.jpg
good1_result.jpg (83.88 KiB) Viewed 7920 times
good2_lines.jpg
good2_lines.jpg (64.06 KiB) Viewed 7920 times
good2_result.jpg
good2_result.jpg (71.42 KiB) Viewed 7920 times
Unfortunately, it turns out even with good representative lines you can still end up with significant wobbling:
wobbling_lines.jpg
wobbling_lines.jpg (79.78 KiB) Viewed 7920 times
wobbling_result.jpg
wobbling_result.jpg (81.5 KiB) Viewed 7920 times
I haven't looked into this issue yet.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
ahmad
Posts: 24
Joined: 28 Dec 2010, 11:26

Re: First build with automatic dewarping

Post by ahmad »

I never cease to be amazed with the work you put into this, Tulon. Thanks once again.

Those 'good' representative lines don't quite seem perfect to my eyes:
Untitled.png
Untitled.png (330.55 KiB) Viewed 7918 times
Is it possible to get ST to calculate all possible representative lines, and then to manually pick two opposed to letting an algorithm do it?
Tulon
Posts: 687
Joined: 03 Oct 2009, 06:13
Number of books owned: 0
Location: London, UK
Contact:

Re: First build with automatic dewarping

Post by Tulon »

ahmad wrote:Is it possible to get ST to calculate all possible representative lines, and then to manually pick two opposed to letting an algorithm do it?
Maybe.
Scan Tailor experimental doesn't output 96 DPI images. It's just what your software shows when DPI information is missing. Usually what you get is input DPI times the resolution enhancement factor.
iso200/24
Posts: 1
Joined: 04 Mar 2014, 00:53

Re: First build with automatic dewarping

Post by iso200/24 »

I logged in to tell you that I was eagerly and patiently awaiting the automatic dewarping function, just to give you some support, and here it is already! This is great and I will be testing it out somewhere this week. Thanks a lot!
Post Reply