Color calculations for and perceptual assessment of computer graphic images

Realistic image synthesis involves the modelling of an environment in accordance with the laws of physics and the production of a final simulation that is perceptually acceptable. To be considered a scientific endeavor, synthetic image generation should also include the final step of experimental verification. This thesis concentrates on the color calculations that are inherent in the production of the final simulation and on the perceptual assessment of the computer graphic images that result. The fundamental spectral sensitivity functions that are active in the human visual system are introduced and are used to address color blindness issues in computer graphics. A digitally controlled color television monitor is employed to successfully implement both the Farnsworth Munsell 100 hues test and a new color vision test that yields more accurate diagnoses. Images that simulate color blind vision are synthesized and are used to evaluate color scales for data display. Gaussian quadrature is used with a set of opponent fundamentals to select the wavelengths at which to perform synthetic image generation. These opponent fundamentals result from an established transform of the fundamental spectral sensitivity functions that minimizes the error of representation for tristimulus values. It is shown that Gaussian quadrature with these opponent fundamentals outperforms Gaussian quadrature with either the fundamental spectral sensitivity functions or the CIE XYZ matching functions. It is also shown that the prioritization that exists between these opponent fundamentals provides further guidance in wavelength selection for realistic image synthesis. A perceptual assessment of computer graphic images is performed by comparing computer generated pictures with a model of a real scene. Under restricted viewing conditions, subjects are unable to distinguish between a view of the real scene and a view of the computer generated picture. This serves as partial proof of the image synthesis process. In another test, the subjects are asked to choose which of two computer generated pictures is a better color match to the real scene. This test independently confirms the superiority of the wavelength selection techniques developed in the thesis.