By Topic

Potential of Multi-Angular Data Derived From a Digital Aerial Frame Camera for Forest Classification

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

2 Author(s)
Koukal, T. ; Univ. of Natural Resources & Life Sci., Vienna, Austria ; Atzberger, C.

The benefits of multi-angular observations for mapping vegetation and their structural and biochemical characteristics are widely recognised. This paper examines the potential of a digital aerial frame camera flown on a standard aerial survey for deriving information on the bidirectional reflectance characteristics of forest and how this information can be used in forest classification. The Rahman-Pinty-Verstraete (RPV) model was fitted to sampled angular observations. For each sample plot at least 10 angular observations were available. The RPV model concentrates the information in a few meaningful parameters and minimizes sensor noise and other perturbing factors. Relying on the fitted model parameters, it is demonstrated that the multi-angular data permits a better discrimination of five forest types as compared to the sole use of spectral information. Compared to the spectral model, the overall classification accuracy (Kappa) increased from 0.702 (0.622) to 0.872 (0.837), when multi-angular data was used. For a reliable approximation of the underlying Bidirectional Reflectance Distribution Function (BRDF) a `balanced' observation geometry is mandatory. In particular, available measurements should cover the backward and forward scattering directions well. Otherwise, the RPV model fits the angular observations well, but the retrieved parameters do not represent the `true' BRDF of the observed target. Under such `unbalanced' conditions a drastic reduction in classification accuracy was observed.

Published in:

Selected Topics in Applied Earth Observations and Remote Sensing, IEEE Journal of  (Volume:5 ,  Issue: 1 )