Quote:
Originally Posted by foobnix
While we are thinking about automatic text blocks recognition I have added to beta ( http://beta.librera.mobi) the simplest algorithm.
My idea is Minimum crop border percent. For my book good result is 12% (custom value),
Crop percent is available only for "Book" mode
Please test is it good or not for your books. I am lazy to add manual crop graphics  |
Sorry, I have not make the test with the last beta version, the one with the 'scroll' mode ... it's take a moment to make the tests, they were done only four hours ago.
I'll try to take the same order to be able to compare with my previous post with the screenshots :
https://www.mobileread.com/forums/sh...&postcount=392.
I've commented some pictures too show the value used by the crop function. The usage of 5 % or 10 % are the major values used.
- Developpement système sous Linux (2 pictures) : you can see that the title of the top on each page is not at the same position;
- Ruby - EffectiveTestingWithRSpec3 (3 pictures) : the last show that the column's size change for 'Part' and the fonts too;
- Linux Magazine (4 pictures);
- ITIL (5 pictures);
- Base de données (2 pictures).
My conclusion : the usage of an percentage works very well.
Thank you for the beta version (Beta-7.8.38-c523) tested.
The new beta version seem to offer what's I need, some different percentage. I'm going to test it. Be patient