The basic motivation behind this project is a highly accurate representation of the human iris, to be injected into a virtual model of the human eyeball. A highly accurate brightness level recording can be easily obtained with a high quality digital camera. Color, however, is an entirely different matter. Photography in the traditional sense entertains all sorts of color inaccuracies, mostly related to the chemical process of development. Digital photography presents gamma and metameric problems, since the exact conditions of the capturing event cannot easily be duplicated. However, the spectral radiance of an object can be captured, utilizing a spectrophotometer and reliable light source. In this research, a priori measurements and analysis of the human iris spectral reflectances are performed. Using a spectroradiometer spectral reflectance samples from human iris are taken and this sample set is analyzed using principal component analysis to give a number of basis functions to reconstruct the original reflectance with sufficient accuracy. A color transformation can be built between the signals from a photometric linear digital camera and the weight coefficients of the eigenvectors. Finally, the spectral reflectance can be derived from the digital counts of the camera giving us a highly accurate representation of a human iris.
Unander, Joseph, "Project Iris: Image reconstruction of the iris spectrally" (2000). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus