The code provided at the url below corresponds the implementation described in:
“Color object recognition based on a clifford fourier transform”.
Jose Mennesson, Christophe Saint-Jean, and Laurent Mascarilla.
In book “Guide to Geometric Algebra in Practice”. Springer, 2011.
For this version,
- you can load images from disk/url so that you can easily try your own images.
- given a (r,g,b) color, you can apply the CFFT : the transform is parameterized by a vector with split the signal into parallel and orthogonal parts (see paper).
- it displays power spectrum for the two parts, but you can easily change the code to display something else.
- you can change interactively (r,g,b) parameters and the transform is recomputed seamlessly.
The code was written in Python 2.7 since i hope to have some time to write a mathematical description of the method with galgebra package.
However, the signal processing part of this notebook should work with Python 3.+.
Give it a try at : https://goo.gl/2QjISn
Source: Email from Christophe Saint-Jean, 22 July 2016, christophe.saint-jean_AT_univ-lr.fr