Date: Abstract
We develop a method which integrates 3-D laser scan data acquired from an aircraft, aerial images, and panoramic video images recorded on the ground for the modeling of the topography (DEM/DTM) and especially natural 3-D objects such as trees and even the shape of the branches. The main tasks are to register the different sensors accurately into the same coordinate system and then reconstruct a model of the environment by fusing the data given by the registered sensors.
Laser scanning produces 3-D geometric data while aerial images contain also radiometric information. Both of them measure the environment from above. The side views are obtained from video images recorded on the ground. Using multiple images, we can measure 3-D data from video and aerial images. On the other hand, laser scanning provides an efficient way to record the geometry from above. The three different sensors complement each other and it is thus important to investigate how to integrate the data provided by the sensors. There are currently several groups working on city modeling from remotely sensed data, but the modeling of trees has not been considered in detail. This makes the project unique.
The results can be exploited in generating simulated sceneries for virtual environments. The registration allows further investigation of the tree modeling and verification of the laser scan data. We will found an international working group on 3-D environment modeling and sensor fusion, where we provide the registered data available to the other participants so that everyone can develop own reconstruction techniques using our data. The project is carried out in cooperation with Dr.-Ing. Eberhard Gülch from Institute of Photogrammetry, University of Bonn, Germany.