Compress scanned handwritten notes with Noteshrink

August 15, 2023

You scan your class notes, but the quality is poor and the file takes up too much space. The solution: Noteshrink.

1. Install Noteshrink

The first thing is to install the Python virtual environment. I am leaving the instructions for Debian 13, although any AI will tell you how to do it on other systems.

On Debian 13, install two packages: python3-venv, imagemagick, and poppler-utils:

$ sudo apt install python3-venv imagemagick poppler-utils

Now create the virtual environment (use the path you prefer):

$ python3 -m venv /home/alejandro/Otros/venv

Activate the virtual environment (replace the path with the one you have chosen):

$ source /home/alejandro/Otros/venv/bin/activate

Finally, install Noteshrink in the virtual environment:

$ pip install noteshrink

2. Configure ImageMagick (if necessary)

On some systems, it may be necessary to modify the ImageMagick security policy to allow handling PDF files:

$ sudo nano /etc/ImageMagick-7/policy.xml

Add the following line inside <policymap>...</policymap>:

<policy domain="coder" rights="read|write" pattern="PDF" />

3. Convert the PDF to images

This article assumes that the notes are in PDF format.

I am going to use the file apuntes-originales.pdf, featuring six pages of my exercises written with a pen.

Convert each page of the PDF into PNG files:

$ pdftoppm apuntes-originales.pdf pagina -png

4. Enhance the images

These two steps are not always necessary:

Experiment with them only if the final result is not satisfactory.

4.1. Increase the contrast

If the colors contrast well with the background, this step is normally not needed. However, when the writing uses light colors, it can be useful to increase the contrast:

$ find . -type f -name "*.png" | parallel convert {} -sigmoidal-contrast 10,200% {}

4.2. Reduce the number of colors

Sometimes it is advisable to reduce the number of colors:

$ find . -type f -name "*.png" | parallel convert {} -level 50%,50% {}

With this, some contrast is gained, the work of Noteshrink is made easier, and the final size is slightly reduced, although it is not an essential step.

5. Compress with Noteshrink

Finally, the time comes to use Noteshrink to process the images and generate an optimized PDF:

$ noteshrink -w -g -O -C -Q -o apuntes-procesados.pdf *.png

6. Result

The processed file will take up much less space than the original file, since:

It reduces the number of colors.
It attenuates or partially removes the background.
It improves the legibility of the handwritten content.

With this process, you can transform heavy and poorly legible scans into lighter and clearer documents.

I have used apuntes-originales.pdf (original file), which was 4.6 MB, to generate apuntes-procesados.pdf (processed file), of just 0.2 MB. The size reduction is spectacular (more than 95%).

Open both files to check the results or consult the following images.

In short, it is worth using Noteshrink.