Application of Remote Sensing and GIS Methods for the Automatic Extraction of Single Trees Based on Digital Aerial Images and Elevation Models

The paper gives a short overview about the existing data base and extraction methods for single tree detection. Forest remote sensing has a long tradition and a variety of methods for single tree extraction have already been developed. Most studies and methods focus either on the analysis of satellite images or airborne laser data and on the extraction of coniferous trees. The automatic detection of deciduous trees is still a great challenge. This paper describes different methods of single tree extraction with focus on the automatic extraction of deciduous trees from aerial imagery. Single trees can be extracted by using aerial true-ortho images and photogrammetrically produced digital surface models as input data and a combination of remote sensing methods and GIS analyses with completeness and correctness over 80 percent. The presented method enables the extraction of tree tops as well as treecrowns for deciduous trees. For the automatically extracted singles trees important attributes like exact position or average crown diameter are calculated and added to the tree objects. The extracted trees can be used for the modeling of trees in virtual environments or for forest area inventories.