Understanding Blur and Model Learning in Projector Compensation
Radiometric compensation enables data projectors to use textured
surfaces such as automobiles, building and theater stages as pro-
jection screens, accomplished by modelling the reflectance char-
acteristic of the surface and inverting it to find the compensation
function. In this paper, we explore the effects of point spread func-
tion / blur of the projector on the performance of existing radiometric
compensation algorithms. Two changes to the existing model are
proposed which help to consider projector blur in model learning.
Proposed changes can be combined with any radiometric com-
pensation strategy to improve its perceptual performance without
increasing the computational complexity.