Image-based Material Characterization for Daylight Simulation Using Illuminance-proxy and Artificial Neural Networks
N. Forouzandeh (TU Delft - Building Physics)
Eleonora Brembilla (TU Delft - Building Physics)
John Alstan Jakubiec (University of Toronto)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
A key aspect of daylight modeling is the definition of material optical properties. Characterization of such properties in existing indoor spaces with current methods is a labour-intensive and time-consuming task, especially in surfaces with considerable visual complexity. Faster and more accurate estimations of such properties will lead to more efficient workflows. Towards this direction, the present work studied the feasibility of using two novel approaches i.e. illuminance-proxy and probabilistic image based material characterization methods for implementation in daylight modeling. These approaches are compared with two common techniques, namely the manual selection from a measured dataset and the use of illuminance/luminance measurements. According to the results, both novel techniques are able to predict spatiallyaveraged Daylight Autonomy, continuous Daylight Autonomy, and Useful Daylight Illuminance in 300-3000 lx range with less than 5% error