Tree Species Classification of Individual Trees in Sweden by Combining High Resolution Laser Data with High Resolution Near- Infrared Digital Images

The aim of this research is to make identification of tree species of individual trees more efficient through combining high resolution laser data with high resolution near-infrared images. Identification of the classes Scots pine (Pinus silvestris L.), Norway spruce (Picea abies L.), and deciduous trees was chosen because these groups are the most important for forest applications in Sweden. Tree species classification is the last step in a method that has the following steps: (1) delineation of individual tree crowns using laser data, (2) estimation of tree height and crown area using laser data, and finally (3) species identification of the delineated tree crowns by adding data from near-infrared digital images. The tests were performed in southern Sweden at the Remningstorp test site (lat. 58°30’N, long. 13°40’E). The laser measurements had a density of seven laser measurements per square meter and the nearinfrared images had a pixel size of 10 cm. On ground, tree position and stem diameter were measured and tree species recorded for trees within a Scots pine dominated, a Norway spruce dominated, and a birch dominated forest stand. Near-infrared images were used for classification. The camera position and orientation of each image was used to map laser generated tree segments to the corresponding pixels in the aerial image. The results indicate that near-infrared images add useful information for tree species classification.