This is a tool to process high-resolution scans with as little user input as possible. It produces decently compressed PDF files with monochrome/color separation and more or less accurate text layers.
In addition to the Python modules listed in
scanproc requires jbig2 to
be installed in your
scanproc.py [-h] [-l LANGS] [-d DPI] [--debug] output [imgs ...]
Usage is simple. You pass a list of
EasyOCR language codes to use
-l (default English only; pass multiple times for more than
one language), the image resolution (required for filter kernel
scaling and correct page size), followed by the output PDF file and
the input image files.