Bidirectional reflectance describes reflectance anisotropy of electromagnetic radiation on different target surfaces. In this thesis, the geometry of bidirectional reflectance was studied and a method for sampling of the phenomenon using multiangle stereocamera images was developed. The objective of the study was to acquire a large amount of directionally-defined data from different natural target surfaces for a BRDF database and modeling.
Surface reflectance anisotropy of targets varies as function of vawelength and viewing and illumination geometries. The phenomenon has a significant role for monitoring of environmental changes or computing large image mosaics using images taken at different times and in varying viewing-illumination geometries. The knowledge of target directional reflectance is also useful when discriminating vegetated targets from each other.
A model was established for sampling of directional effects from multiangle images. The intersection of digital terrain model and a sample vector was solved in order to provide a single target point for sampling. That was carried out by an iterative algorithm. Viewing and illumination geometry of the sample was defined in cases of flat and oblique surfaces. The images were classified in order to group the samples adequately.
Using the developed sampling process, a large amount of directionally-defined samples was acquired. Taking the surface orientations into account, wider angular range of the sample geometry was attained. The applicability of the derived data were tested by studying the bidirectional reflectance of six agricultural and forest targets. The multiangular image data, developed sampling methods and bidirectional dataset were discovered to be feasible for investigations of bidirectional effects of natural targets.
Bidirectional modeling and image correction are proposed to be studied further. Further, the atmospheric effects should be taken into account in a more detailed manner.