Color heterogeneity of building surfaces: lean image processing approach for visible reflectance characterization performance

Daylighting availability and uniformity depend on the interior surface reflectance (ρ). Currently, ρ is obtained through suggested reference values (Illuminating Engineering Society 2012; CIBSE 2015; CIBSE/SLL 2011; CIBSE/SLL 2005), laboratory tests (ASTM E 903 (2012) describes a standardized procedure requiring a calibrated instrument and an as-built surface sample), or by on-site measures (based on luminance and illuminance differences). Novel methodologies compute it by integrating image processing and/or photometry, applied on false colour or HDR images. A simple procedure is needed for accurately assess, even in preliminary design phases, the reflectance of heterogeneous surface areas for new and historical buildings. In fact, heterogeneous surfaces (colour, texture, composition, ageing) difficult the accurate estimation of a representative reflectance value (?̅?) for building simulation, leading to daylighting performance deviation. This work presents a methodology, based on a per-pixel colour reflectivity (ρcol) evaluation, to easily acquire an approximate value of the surface visible reflectance (?̅?vis). This approach gives a more global ?̅?vis of all surface components, aiming to improve the accuracy of the modelled daylighting analysis. Additionally, the procedure is experimented over a sample reference test room.