The Generation of 3D Tree Models by the Integration of Multi-sensor Data

Abstract

Three-dimensional tree modeling is an important task in the management of forest ecosystems. The objective of this investigation is to reconstruct 3D tree models using LIDAR data and high resolution images. The proposed scheme comprises of three major steps: (1) data preprocessing, (2) vegetation detection, and (3) tree modeling. The data preprocessing includes spatial registration of the airborne LIDAR and high resolution images, derivation of the above ground surface from LIDAR data, and generation of a spectral index from high resolution images. In the vegetation detection, a region-based segmentation and knowledge-based classification are integrated to detect the tree regions. Afterwards, the watershed segmentation is selected to extract the tree crown and heights. In the last step, we use the tree height, tree crown and terrain information to build up the 3D tree models. The experimental results indicate that the accuracy of the extracted individual tree is better than 80%, while the accuracy of the determined tree heights is about 1m.