Abstract

All the different approaches taken for spectral data acquisition can be narrowed down to two main methods; the first one is using spectrophotometer, spectroradiometer, hyper- and multi- spectral camera through which the spectra can be most probably attained with a high level of accuracy in a direct manner. Nonetheless, the price at which the spectra are acquired is very high. However, there is also a second approached in which the spectra are estimated from the colorimetric information. The second approach, even though it is very cost efficient, is of limited level of accuracy, which could be due to the methods or the dissmiliarity of learning and testing samples used. In this work, through looking upon the spectral estimation in a different way, it is attempted to enhance the accuracy of the spectral estimation procedures which is fulfilled by associating the spectral recovery process with spectral sensitivity variability present in both different human observers and RGB cameras.

The work is split into two main sections, namely, theory and practice. In the first section, theory, the main idea of the thesis is examined through simulation, using different observers’ color matching functions (CMFs) obtained from Asano’s vision model and also different cameras’ spectral sensitivities obtained from an open database. The second part of the work is concerned with putting the major idea of the thesis into use and is comprised of three subsections itself. In the first subsection, real cameras and cellphones are used. In the second subsection, using weighted regression, the idea presented in this work, is extended to a series of studies in which spectra are estimated from their corresponding CIEXYZ tristimulus values. In the last subsection, obserevers’ colorimetric responses are simulated using color matching. Finally, it is shown that the methods presented in this work have a great potential to even rival multi-spectral cameras, whose equipment could be as expensive as a spectrophotometer.

Publication Date

7-5-2017

Document Type

Thesis

Student Type

Graduate

Degree Name

Color Science (MS)

Advisor

Mark D. Fairchild

Advisor/Committee Member

Susan Farnand

Campus

RIT – Main Campus

Share

COinS