Abstract

On the basis of characteristic vector analysis (CVA) and multiple regression analysis (MRA), a method of characteristic vector regression analysis (CVRA) is proposed for the data reduction of the spectra database. The high correlation among the spectra in the spectra database ensures the feasibility of the method. Studies of Gaussian functions and spectra of groundwaters, as well as spectra of soils and plants, with evidence of a high reduction ratio and good retrieval precision, show the effectiveness of the method. Better retrieval precision can be achieved by combining the characteristic vectors with the mean spectrum for the reconstruction of the original spectra. The method can be extended to perform spectral classification and spectral background correction.

Application of Characteristic Vector Analysis to Photographic and Optical Response Data*

J. L. Simonds
J. Opt. Soc. Am. 53(8) 968-974 (1963)

Analysis of optical spectra by Fourier methods: filtering and least-squares regression in reciprocal space

D. E. Aspnes and H. Arwin
J. Opt. Soc. Am. 73(12) 1759-1764 (1983)

Statistical analysis of a database of absorption spectra of phytoplankton and pigment concentrations using self-organizing maps

Aymeric Chazottes, Annick Bricaud, Michel Crépon, and Sylvie Thiria
Appl. Opt. 45(31) 8102-8115 (2006)

Ocean color spectra measured off the Oregon coast: characteristic vectors

James L. Mueller
Appl. Opt. 15(2) 394-402 (1976)

A Computer Method for the Automatic Reduction of Spectroscopic Data

Earle F. Ditzel and Lorrain E. Giddings
Appl. Opt. 6(12) 2085-2090 (1967)