This paper describes an approach for building detection from multispectral
aerial images and lidar data by combining the results derived from statistical
and neural network classifiers, which offer complementary information, based
on Dempster-Shafer Theory of Evidence. Four study areas with different
sensors and scene characteristics were used. First, we filtered the lidar point
clouds to generate a Digital Terrain Model (DTM), and then the Digital Surface
Model (DSM) and the Normalised Digital Surface Model (nDSM) were
generated. After that a total of 25 uncorrelated feature attributes have been
generated from the aerial images, the lidar intensity image, DSM and nDSM.
Then, three different classification algorithms were used to detect buildings from
aerial images, lidar data and the generated attributes. The classifiers used
include: Self-Organizing Map (SOM); Classification Trees (CTs); and Support
Vector Machines (SVMs). The Dempster-Shafer theory of evidence was then
applied for combining measures of evidence from the three classifiers. A
considerable amount of the misclassified building pixels were recovered by the
combination process.