Gone are the days when you needed to go to the library for a book. Now you can download the book electronically, load it on your e-reader or tablet, and start enjoying it. But all electronic books are not equal. Particularly infuriating are electronic books that are actually scanned images of old print books. Scanned images of old books, which typically come in PDF format, are difficult to read on a black-and-white E Ink screen, where fading text and yellowing pages appear as blurs, blotches, and dark-gray backgrounds.

Luckily, you can clear up that blurry scanned image with a few tricks from ImageMagick. This article describes a method you can use to spruce up a scanned electronic book. Note: If you obtained the book from a lender or through another vendor, be sure the license supports this type of file manipulation.

Getting Started

I needed a copy of a sociology book, A Place on the Corner, by Elijah Anderson; the only place I could find it in electronic format was The Internet Archive. They had a scanned copy, so I loaded it on my Sony DPT-RP1/B e-reader, but the text was difficult to read (Figure 1). Dark spots appeared on the page, and the text was just a bit darker than the background, with poor contrast (Figure 2).

Figure 1: The scanned PDF book as it displays on a Sony DPT-RP1/B e-reader's 13" E Ink screen.

Figure 2: The contrast between the text background is insufficient, impeding readability.

Hoping to find a better view, I installed the pdfimages package on my Ubuntu 18.04 system:

sudo apt install pdfimages

pdfimages is used to extract images from PDF files and save them as Portable Pixmap (PPM), Portable Bitmap (PBM), or another format. Since the scanned pages in the book are image files, all you need to do is drop the PDF into an empty folder and enter the following command:

pdfimages A_place_on_the_corner.pdf

This command will generate a large amount of PPM and PBM image files (Figures 3 and 4). The PPM files contain all the shadows and imperfections of the scanned pages, and the PBM files are a negative, clean, white-on-black version. Now all you need to do is reverse the colors in the PBM negatives. You don't need the PPM files, so you can remove them:

rm *.ppm

Figure 3: This is how a PPM file extracted from a scanned book will eventually look.

Figure 4: This is a PBM file extracted from the scanned book with pdfimages utility.

To maintain quality, convert all remaining PBM files into PNG format using mogrify, which is part of the ImageMagick package. (See also the "More Memory?" box for using ImageMagick with large files.) First be sure to install ImageMagick:

sudo apt install imagemagick

01 <policy domain="resource" name="memory" value="2256MiB"/>
02 <policy domain="resource" name="map" value="512MiB"/>
03 <policy domain="resource" name="width" value="16KP"/>
04 <policy domain="resource" name="height" value="16KP"/>
05 <policy domain="resource" name="area" value="128MB"/>
06 <policy domain="resource" name="disk" value="20GiB"/>

mogrify is used to manipulate graphic files: rotate, crop, flip, blur, and join. You can also use mogrify to convert from one format to another. Search for all PBM files in the folder and use mogrify to convert them to PNG format automatically:

find -name '*.pbm' -print0 | xargs -0 -r mogrify -format png

Now that the pages of the book are in PNG format, you don't need the PBM files anymore, so you can delete them:

rm *.pbm

convert is another command-line interface tool shipped with ImageMagick that does about the same thing as mogrify, but it is also able to invert the colors of an image file. The PNG files are currently in negative, so you can convert them to look like regular book pages using the -negate attribute of convert. Output the results as JPG files to keep the file sizes low:

ls -1 *.png | xargs -n 1 bash -c 'convert "$0" -negate "${0%.png}.jpg"'

You can now delete the PNG files and inspect the remaining JPG images. Should you feel they need more contrast, you can use the -level argument of the convert command to bulk-modify the contrast of your files:

ls -1 *.jpg | xargs -n 1 bash -c 'convert "$0" -level 60 "${0%.jpg}.jpg"'

Replace 60 with a value ranging from 1 to 100, with a higher value for more contrast, or reduce the value for less contrast. The next step is to insert the now-cleaned and easily-readable scanned pages into a PDF file. Should you wish to keep the resulting PDF as small as possible, you can batch-resize the JPG files beforehand to 50 percent of their actual size using mogrify:

ls -1 *.jpg | xargs -n 1 bash -c 'mogrify "$0" -resize 50 %"${0%.jpg}.jpg"'

All the files were extracted from the scanned PDF book and numbered in order, and the file names haven't changed through all the conversion steps, so the entire book is ready to insert, page by page, back into a PDF. You can easily add the converted pages back in with:

convert *.jpg a_place_on_the_corner-purified.pdf

The result is a PDF file that has crisp black text on a white background, and the difference is noticeable (Figure 5). You can load it on your e-reader or tablet and read it without eye strain.

Figure 5: The output PDF file looks better on the E Ink screen and is now easily readable.

Conclusion

Using this PDF purifying method, you can get rid of pinkish or yellow backgrounds from scanned documents. Figure 6 shows a comparison of the purified text with the original version of the book. This method will also help you remove other artifacts of the scanning process, such as the faint text lines that appear from the text on the other side of the scanned page, as well as dirty fingerprint marks, light coffee stains, or pencil marks.

Figure 6: A side-by-side comparison of the purified page with the original. Except for a tiny black spot that remained from an ink drop on the original page, the converted version is black and white, and the faint outline of text from the other side of the printed page is gone.