SIMULATION OF WHITE-LIGHT ADAPTATION CHARACTERISTICS WITH USE OF NONLINEAR NEURAL PRINCIPAL COMPONENT ANALYSIS

Citation
K. Mantere et al., SIMULATION OF WHITE-LIGHT ADAPTATION CHARACTERISTICS WITH USE OF NONLINEAR NEURAL PRINCIPAL COMPONENT ANALYSIS, Journal of the Optical Society of America. A, Optics, image science,and vision., 14(9), 1997, pp. 2049-2056
Citations number
45
Language
INGLESE
art.tipo
Article
Categorie Soggetti
Optics
ISSN journal
1084-7529
Volume
14
Issue
9
Year of publication
1997
Pages
2049 - 2056
Database
ISI
SICI code
1084-7529(1997)14:9<2049:SOWACW>2.0.ZU;2-5
Abstract
We use a nonlinear neural principal component analysis to approximate and simulate white-light adaptation characteristics of opponent color signals and an achromatic signal. The algorithm used was derived from Sanger's generalized Hebbian algorithm [Neural Netw. 2, 459 (1989)] wi th use of nonlinearities in short-wavelength cone outputs. The princip al components can be interpreted as one achromatic and two opponent co lor signals. Simulation examples show that the algorithm can approxima te opponent color signals and adaptation characteristics for the red-g reen signal that closely resemble those reported in the literature. Th e model used incorporates nonlinear models of the opponent mechanism w ith white-light adaptation characteristics and allows a nonlinear adap table interaction between opponent mechanisms. (C) 1997 Optical Societ y of America.