Scan Tailor "Enhanced"

[pejuko]

Hi,

I've been working on some new ScanTailor features which are in git repository but they are not included in any ScanTailor release yet. I've prepared this development release that you can test and use it if you find it useful.

This release: Tiff compression option allows to disable compression in tiff files or select one of other standard compression methods (lzw, deflate, packbits, jpeg).

Deviation patch enables highligting of different pages. Highlighted in red are pages from Deskew filter with too hight skew, from Select Content filter pages with different size of content and in Margins filter are highlighted pages which does not match others.

Automargins patch allows keep page content on original place. In the Margins step you can choose from Auto, Manual (default) and Original mode.
The manual mode is the original one. Auto mode try to decide if it is better to align page top, bottom or center. Original mode keeps page on their vertical original position.

Pagedetect patch allows detect page in black margins and enables keeping original page layout or switch off page content detection. To detect and keep original page layout from command line try:

Code: Select all

scantailor-cli --disable-content-detection --enable-page-detection --enable-fine-tuning --margins=0 images outdir/

Or you can let the content detection enabled and enable also auto margins:

Code: Select all

scantailor-cli --enable-page-detection --enable-fine-tuning --enable-auto-margins images outdir/

Matchlayout patch adds --match-layout options to cli version which detects atypical pages and does not match their size with others automaticaly.

Pictureshape patch adds option for mixed pages to choose from free shape and rectangular shape images. This patch does not improve the original algoritm but creates from the detected "blobs" rectangular shapes and the rectangles that intersects joins to one.

For windows I've compiled executable installation package. Other platforms have to compile from prepared source code which you can download from the same place or merge them yourselfs from git.

You can download the files from https://sourceforge.net/projects/scantailor/files/scantailor-devel/enhanced/

Latest version: 20120503
This post updated 5/3/2012

Petr Kovar

[Misty]

Hi, Petr,

Is this release stable enough for general use? Is there a git tag I can use to pull the code in this release? If so, I was going to add this as a build option in Homebrew so Mac users can easily try out the development release.

[Tulon]

Hi Petr,

I am glad you still hang around. Your post reminded me I promised you to review the automargins code, which I never did. Sorry about that. I rarely work on ST these days, as I got tired of that. I hope you understand.

I do support people releasing their own versions of ST. After all, the Free Software / Open Source thing is all about being able to do just that. Keep up the good work!

[daniel_reetz]

Petr,

Recognize you from the Scan Tailor dev-list. Thanks for setting this up and sharing with the community here. Hopefully we can help you build for all platforms.

[eL_PuSHeR]

Thank you very much for this release. Your hard work is appreciated.

[jimboh]

Would you post a screenshot of Pictureshape in action?

[pejuko]

Would you post a screenshot of Pictureshape in action?

There is an example:

The original "free shape" mode:

scantailor-freeshape.cut.400.jpg (28.61 KiB) Viewed 8026 times

and "rectangular shape" mode:

scantailor-rectangularshape.cut.400.jpg (23.98 KiB) Viewed 8026 times

Both screen-shots are taken with Picture Zones enabled.

[pejuko]

I do support people releasing their own versions of ST. After all, the Free Software / Open Source thing is all about being able to do just that. Keep up the good work!

Would it be possible to create on SF new folder for such releases?

[Tulon]

Yes, sure. I just gave you file release permissions. Now you should be able to create folders and upload files.

[pejuko]

Is this release stable enough for general use? Is there a git tag I can use to pull the code in this release? If so, I was going to add this as a build option in Homebrew so Mac users can easily try out the development release.

Misty,

I'm using ScanTailor mostly on linux and with the cli version I've processed automatically thousands of pages with very good result.

Files are now on SourceForge in subdirectory:
https://sourceforge.net/projects/scantailor/files/scantailor-devel/enhanced/
so it should be more comfortable for packagers to download them now.

Instead of pulling from git I recommend to use the .tar.gz package. It's merged from four branches and there are conflicts.