Regular paper

[OPTICAL REVIEW Vol. 16, No. 2 (2009) 81-90]
© 2009 The Optical Society of Japan

Principal Component and Sampling Analysis of Color Spectra

Juha LEHTONEN^*, Jussi PARKKINEN, Timo JAASKELAINEN¹, and Alexei KAMSHILIN²

Department of Computer Science and Statistics, University of Joensuu, P. O. Box 111, FIN-80101 Joensuu, Finland
¹Department of Physics and Mathematics, University of Joensuu, P. O. Box 111, FIN-80101 Joensuu, Finland
²Department of Physics, University of Kuopio, P. O. Box 1627, FIN-70211 Kuopio, Finland

(Received September 9, 2008; Accepted January 26, 2009; Revised January 15, 2009)

We studied the effect and relation between the optimal sampling interval and the optimal number of principal component analysis eigenvectors for color spectra. The effect of different type of light sources and illuminants are included in the study. Several reflectance spectra sets under different real light sources and standard illuminants were used as a test set. The needed sampling interval and number eigenvectors with different light sources and illuminants are found. The used bandwidth in measurements causes effect to the resulted optimal sampling interval, which is also shown.

Key words: color spectrum, light source, principal component analysis, sampling interval

© 1994-2009 The Optical Society of Japan (An Affiliate of the Japan Society of Applied Physics)
Produced, Developed, and Maintained by The Optical Society of Japan (An Affiliate of the Japan Society of Applied Physics)
Printed in Japan by Komiyama Printing Co., Ltd.

mail to Editorial Office, OPTICAL REVIEW