PROJECT NOTES / DATA
Eigenfaces · PCA Exploration
Finding structure in pixels.
A linear algebra project using singular value decomposition to compute eigenfaces and reconstruct images with fewer components.
- Python
- NumPy
- Matplotlib
- Pillow
- SVD
THE CONTEXT
A closer look.
The problem
High-dimensional face images can be represented with a smaller set of principal components. This project explores how linear algebra produces that reduced representation.
What I built
The public project loads and resizes PGM images, computes a mean face, applies PCA through SVD, and reconstructs a selected image using a chosen number of components.
HOW IT FITS TOGETHER
The architecture.
- 01
Load PGM images and resize to a consistent shape.
- 02
Compute the mean face and center the image data.
- 03
Use SVD to obtain principal components.
- 04
Reconstruct an image using a reduced component set.
ENGINEERING DETAILS
Decisions that matter.
Numerical foundations
NumPy handles the array operations and singular value decomposition used to derive eigenfaces.
Visual exploration
Pillow handles images and Matplotlib supports visualization of the reconstructed outputs.
Public Eigenfaces README. Presented as a linear algebra assignment; no recognition accuracy or deployment claims are made.
UP NEXTProxmox Homelab