Take jpgs, crop and adjust them, and then pass them through Tesseract. https://github.com/tesseract-ocr/tesseract
Install the basics:
git clone https://github.com/jhub95/ocrscantools
apt update
apt install tesseract-ocr libthread-queue-any-perl libmouse-perl libconfig-general-perl liblist-allutils-perl imagemagick libpath-tiny-perl libsys-cpuaffinity-perl libfindbin-libs-perl python-pil python-reportlab libtest-differences-perl
installing the tesseract new word database for Turkish:
sudo cp tesseract-train/hasat_tur.traineddata /usr/share/tesseract-ocr/4.00/tessdata/
sudo cp tesseract/hasat_txt /usr/share/tesseract-ocr/4.00/tessdata/configs/
Test to see if the programs need recompilation for your OS:
./htmlout
If that produces errors than recompile like:
apt install libtesseract-dev dpkg-dev libopencv-dev
g++ -std=c++0x -O3 htmlout.cpp -ltesseract -llept -o htmlout
g++ -O3 -o detect_page detect_page.cpp -lopencv_core -lopencv_objdetect -lopencv_highgui -lopencv_imgproc -lopencv_imgcodecs
Change the lib include path:
in the newly downloaded convert_book.pl file change line 33 to match
the username of the location you installed ocrscantools.
--
i.e. the default is: use lib '/home/bookscanner/ocrscantools/lib';
but you installed the ocrscantools in /home/justin/ocrscantools.
Just replace 'bookscanner' with 'justin', then save the file.
I don't think this is still required, but you can build the patched version (ver 3 - only tested on ubuntu 14) like:
apt-get source tesseract
get pdf_background_image.patch
cd tesseract-dir
patch -p0 < ../pdf_background_image.patch
dpkg-buildpackage