Remote sensing of urban materials by their bidirectional reflectance distribution function

University essay from Luleå/Space Science, Kiruna

Abstract: Remote sensing systems exploit optical properties to gain information about
a target. In the majority of the cases, this is done by exploiting the
spectral domain. Only little effort is spend on looking for alternatives.
The idea, that is investigated in this study, is to take advantage of the
bidirectional reflectance which may carry information about the type of
material, of which the target is composed. This method may be interesting
especially for urban regions, where targets with different inclinations and
orientations are distributed on a small area. A camera, which views two
targets of the same material receives different fluxes according to the
inclination of each target. That can disturb a classification process that
is based on spectral information only, and does not take into account
bidirectional reflectances. As for spectral data, bidirectional reflectance
data is available for many types of materials. However, data sets are
mostly restricted to incidence and reflection zenith angles of 60°.
Therefore, to start with, measurements are conducted to acquire data which
covers the whole upper hemisphere of a sample, i.e. a solid angle of 2pi.
For this first phase, the goniometer available at ONERA in Toulouse is
used, which, as well, is restricted to 60° zenith angle. An inclination
wedge is used in order to measure angles up to a zenith angle of 80°.
Limitations of this method are found. Seven real-world urban samples have
been measured respecting the limitations of the measurement method. They
represent well the diversity of materials which can be found in a city.
Surprinsingly, for some of the samples, measurements show an increase in
reflectance in the forward direction for grazing zenith angles. Three of
the seven data sets are fitted to recent successful BRDF models. A model
which is able to successfully describe BRDFs of urban materials has not
been found. However, the model developed by R.L. Cook and K.E. Torrance
describes the above mentioned effect of increasing reflectance for
grazing angle surprisingly well. It is considered to be a good candidate
for further investigations on the way to establish a classification
algorithm which is based on bidirectional reflectances.