cs781 - Colour for Computer Graphics - Winter 2012

Course Notes

Lecture 15 - Linear Models of Reflectance

And Now from our Sponsor

1. Projects
2. Last three lectures
3. Talk on flicker hallucinations
   Dr. Bard Ermentrout (Computational Biology, University of Pittsburgh,)
   Flicker Hallucinations: Faraday waves in the brain
   3.30 p.m., Tuesday, 6 March
   MC 5158.

Linear Models of Colour

Where did it come from

• remote sensing
• paint analysis

Principal Components Analysis

Given a very large sample of data, how can you summarise it?

• e.g. spectral power distributions from many sources of light
• PQR decomposition

Weaknesses

1. The usual least squares problems
   • and there is no well-accepted way of doing other metrics
   • Why? Because of the other problems
2. Sample dependence
   • You have to redo it for each new sample
3. There is usually not a well-defined break point

Possible Remedies

1. We know the dimension
   • e.g. colour display

PCA of Coloured Surfaces

PCA of Coloured Light

Return to:

• Bill Cowan's Winter 2012 cs781 lecture note page
• Bill Cowan's Winter 2012 cs781 page
• Bill Cowan's cs781 page
• Bill Cowan's teaching page
• Bill Cowan's home page