Principal component analysis, linear algebra and so on..

meh

Hello!

My question is very simple: starting from a PImage with its own pixel array, it is possible to implement image reduction methods like Principal Component Analysis or SVD - that make a large use of linear algebra - without installing any additional library? I am quite confused about the PMatrix role: there is a chance to create new matrices (and maybe, load the pixel array in a matrix)?

Chrisir

it's possible

PMatrix has nothing to do with that.

PMatrix is the positioning of the coordinate system in 3D space

of course, the images come in a kind of grid in the image

for images and pixels see tutorial

https://www.processing.org/tutorials/pixels/

Chrisir ;-)

meh

Ok, I see. But without PMatrix is impossible to do think of doing complex matrix-like operations on my PImage (and that becomes a big obstacle when it comes to eigenvectors and eigenvalues).

T_D

Here is an older PCA-demo built with processing 1.5.1:

https://www.openprocessing.org/sketch/101449

Shouldn't be too hard to port to Processing 3.

meh

I already saw it - a nice one. Unfortunatly I am working for a course task - and it is requested to not to use external libraries

KevinWorkman

Please link between crossposts so people don't waste their time giving you advice you've already received. this question has also been posted here: https://stackoverflow.com/questions/47939486/pca-or-svd-in-processing

meh

sorry about that. The stackoverflow thread has been deleted.