Skip to content

A mathematical model of handwriting, as described by Daniel Martín-Albo.

Notifications You must be signed in to change notification settings

andrew-healey/sigma-lognormal

Repository files navigation

A model of human handwriting

This project is the first public implementation of the sigma-lognormal model.

The sigma-lognormal model is a mathematical model that elegantly describes human handwriting.

It posits that handwriting and mouse movements are composed of a series of a few overlapping handstrokes, which are simple circular arcs with the following speed profile:

image

This repository implements a "sigma lognormal handstroke extractor" for human handwriting--it breaks down a piece of handwriting into a few handstrokes.

You can find the papers on which this implementation is based in the papers folder, along with annotations and derivations of my formulas.

Example of the extractor in action:

Usage

Installation

pip install -r requirements.txt

Running

Run the Jupyter Notebook demo.ipynb.

To-Do

  • Make unified plotting/animation API.
  • Decrease drift on a battery of signatures.
  • Add web reconstruction interface
  • Use PyTorch to fine-tune extracted parameters with gradient descent.
  • Add mouse movement-specific extractor.
  • Possibly add a ghost-cursor-style mouse movement generator.

About

A mathematical model of handwriting, as described by Daniel Martín-Albo.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published