Towards extended material probing using TLS and partial bidirectional reflectance distribution functions

Terrestrial laser scanning (TLS) intensity values can provide meaningful information about scanned surfaces. Most reported TLS-based material probing attempts rely on deriving a reflectance constant per individual object surface, assuming that this reflectance primarily depends on the surface material properties. However, surface reflectance cannot be fully represented by a scalar value. Each surface exhibits complex reflectance patterns that can be represented by a characteristic bidirectional reflectance distribution function (BRDF), which encodes further material properties, surface roughness, and microstructure. While TLS data do not allow retrieval of the full BRDF, they do permit derivation of a special case thereof, corresponding to the configuration where illumination and observation directions coincide. We refer to this as the material-dependent angle-of-incidence compensation function (AOICOF). We demonstrate that estimating the AOICOF from point clouds enables material identification. We achieve this by compensating the intensities for range dependence, segmenting the point clouds of a scene into surface patches, jointly processing patches with common material properties, and estimating AOICOF for each such group of surface patches. The AOICOFs estimated from the scans of the scene are then compared to a catalogue of reference functions obtained from laboratory measurements of various materials. We evaluate the agreement between in-situ and laboratory data by comparing their AOICOF and representative reflectance constants. Results demonstrate that obtained AOICOFs align closely with laboratory references, effectively allowing the distinction of material classes. While the shape of the AOICOF serves as a filter for narrowing down potential materials, we find that combining it with a representative reflectance constant helps distinguish between materials with similar AOI dependent reflectance response. With these results, we demonstrate that reusing radiometric compensation functions enables extended material probing, providing a more comprehensive characterization of surfaces than a reflectance constant alone.