A psychophysical investigation of global illumination algorithms used in augmented reality

摘要

The overarching goal of this research was to compare different rendering solutions in order to understand why some yield better results specifically when applied to rendering synthetic objects into real photographs. A psychophysical experiment was conducted in which the composite images were judged for accuracy against the original photograph. In addition, iCAM, an image color appearance model was also used to calculate image differences for the same set of images. Conclusions obtained included the effect of global illumination on the accuracy of the final composite rendering. Also, it was discovered that the original rendering with all of its artifacts is not necessarily an indicator of the final composite image's judged accuracy. Finally, initial results show promise in using iCAM to predict a relationship similar to the psychophysics, which could eventually be used in-the-rendering-loop to achieve photorealism with minimized computation.