Purchase this article with an account.
Noriko Yata, Tomoharu Nagao, Keiji Uchikawa; A categorical color perception model using artificial neural network. Journal of Vision 2005;5(12):96. doi: https://doi.org/10.1167/5.12.96.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The purpose of this study is to get a model that can operate similarly to human beings categorical color perception and color constancy. Color of object is not determined only by reflection spectrum from the surface of the object alone. It is greatly affected by ambient environmental conditions and depends upon color constancy. Mechanism of color constancy, however, is not explained in detail. Therefore, it is difficult to acquire cognition of categorical color name of objects under different illuminations.
For this end, the relationship between chromaticity of color chips under different illuminations and categorical color names due to the color chips under the illuminations ware learned by an artificial neural network. The categorical color names are products of a categorical color naming experiment on four normal color sensation men that use OSA color chips and three illuminants. The network is trained using the error Back-propagation algorithm that is the most popular learning algorithm for feed-forward neural networks. Experimental results show that the obtained neural network has the similar characteristics to those of human beings vision system.
This PDF is available to Subscribers Only