Specifying option --oversample tends to introduce upsampling in rendering
by rasterizing page to an higher DPI.
This upsampling improves OCR results, but a correct choice of interpolation
method can increase even more the OCR quality.
Ghostscript seems to use a nearest interpolation as default choice for pdf.
This method doesn't average new introduced pixels with original pixels
resulting in an almost similar image but with more pixels.
Providing -dInterpolateControl=-1 force switching interpolation on.
In this commit the above option is passed to all ghostscript rendering
calls.
After testing, rendering a page at same DPI with interpolation
enabled does not introduce significant time overhead.
time (repeat 40 gs -dQUIET -dSAFER -dBATCH -dNOPAUSE -sDEVICE=png16m \
-dFirstPage=1 -dLastPage=1 -r100.000000x100.000000 \
-dInterpolateControl=-1 -o /dev/null -dAutoRotatePages=/None -f pzII.pdf)
7,66s user 0,33s system 99% cpu 8,012 total
time (repeat 40 gs -dQUIET -dSAFER -dBATCH -dNOPAUSE -sDEVICE=png16m \
-dFirstPage=1 -dLastPage=1 -r100.000000x100.000000 \
-o /dev/null -dAutoRotatePages=/None -f pzII.pdf)
7,42s user 0,39s system 99% cpu 7,808 total
Ghostscript interpolation control reference:
https://www.ghostscript.com/doc/current/Use.htm
PDFs are quite likely to have a lot of pixels, e.g. large high resolution scans.
250 MP is a page of A0 sized paper scanned at 400 DPI,
should be enough in most cases.
Tesseract is now included better thresholding (binarization) in v5. Users that have
thresholding issues should try that first. If we find further problems
this can be brought back as a plugin.
When compiled without pybind11 2.8.0, pikepdf supplies a shim to implement
pikepdf._ObjectMapping.values() which has subtly different semantics
from a true dict-like objects; in particular it supports
next(objectmap.values())
where a standard dict requires
next(iter(objectmap.values()).
pybind11 2.8.0 now implements .values() properly, meaning some misuses of
protocol in ocrmypdf fail.
If pybind11 < 2.8.0, pikepdf will
continue to offer its shim. If pybind11 >= 2.8.0, pikepdf does not add its shim.
Consequently no changes were needed in pikepdf.
Closes#843
Error is:
TypeError: from_buffer() got a 'memoryview' object, which supports the buffer interface but cannot be rendered as a plain raw address on PyPy
PyPy is happy to access a bytes() copy of the memoryview.
James R. Barlow
Tristan PorteriesandGitHub
fedeliallalineaandGitHub