Ubuntu – Convert an image from grayscale to binary

convertimage processingimagemagickpdf

I was wondering how to convert an image from grayscale to binary. The image is a one-page document in pdf format.

If I use Imagemagick's convert 1.pdf 1.eps to convert it to eps or other formats such as tif, I find the image becomes blurry. How can I preserve the clarity of the image?

Thanks!

Best Answer

Hm, it seems like that image is stored into the PDF as a bunch of long small "strips" of the complete image. It also seems like there is a filter or set of filters applied in the PDF that makes the image to be shown as b/w instead of the graylevels that it really is. You can open it in Inkscape, for example, to see some of this for yourself, as it seems like these filters aren't imported (I guess the same happened when you converted to PS or TIFF).

Turning it from a grayscale to a bitmap will probably require a bit of playing with contrast & such to get an optimal result (the scan isn't very hi-res to start with...).

You can import the PDF into The GIMP and play with various filters and colour-options, or maybe use similar functions in imagemagick to create a bitmap version. It's difficult what filter/setting will give the best results...

Related Solutions

PDF – How to Batch Convert an Image to a PDF

One workaround is to split the image generation and the PDF conversion. First convert the images via convert to A4@300dpi (i.e. 3508x2479), then use sam2p to convert them to PDF and then use sam2p_pdf_scale to convert them to A4.

convert -rotate "90>" -scale 3508x2479 -border 64x64 -bordercolor white in.png out.png
sam2p out.png out.pdf
sam2p_pdf_scale 595 842 out.pdf

Edit: A more complete script:

#!/bin/sh

A4_WIDTH=2479
A4_HEIGHT=3508

H_MARGIN=64
V_MARGIN=64
WIDTH=$((${A4_WIDTH} - ${H_MARGIN} * 2))
HEIGHT=$((${A4_HEIGHT} - ${V_MARGIN} * 2))

for i in "$@"; do
    TMP="/tmp/$(uuidgen).png"
    echo "$i"
    convert \
        -rotate "90>" \
        -scale "${WIDTH}x${HEIGHT}" \
        -border "${H_MARGIN}x${V_MARGIN}" -bordercolor white \
        -gravity center \
        -extent "${A4_WIDTH}x${A4_HEIGHT}" \
        -gravity center \
        -font helvetica -pointsize 80 \
        -fill white -draw \
        "push graphic-context
         translate $((${A4_WIDTH}/2 - 160)), 0
         rotate 90
         text -2,-2 '$i'
         text -2,2 '$i'
         text 2,-2 '$i'
         text 2,2 '$i'
         pop graphic-context
    " \
        -fill black -draw \
        "push graphic-context
         translate $((${A4_WIDTH}/2 - 160)), 0
         rotate 90
         text 0,0 '$i'
         pop graphic-context
    " \
        "$i" "$TMP"
    sam2p "$TMP" "${i}.pdf"
    sam2p_pdf_scale 595 842 "${i}.pdf"
done

# EOF #

Ubuntu – Flattening PDF transparency

One way around bad rasterization is to manually convert your PDF document to postscript because when you invoke CUPS through evince or whatever gui, it will be converted anyways.

You can try pdftops -r 1200 document.pdf to increase the resolution from the low default of 300, and then print the resulting postscript. However, with the printers at work this did not yield a significant improvement. Regular text on pages with and without transparent images still looked different -- blurry vs. clear.

So I ended up using Acrobat Professional (on a windows machine) to flatten the transparent images used in my latex document, which gave optimal results. Check that Transparency Flattener Preset is at [High Resolution] in the advanced print setup. My colleague mentioned that one could perhaps FTP a pdf file directly to a modern laserjet network printer, bypassing the postscript conversion. I have not tried this.

Would be great to have linux-only solution to this issue.

Best Answer

Related Solutions

PDF – How to Batch Convert an Image to a PDF

Ubuntu – Flattening PDF transparency

Related Question