Identification and classification of scanned target in forest based on hierarchical cluster

In order to avoid the incorrect operation of the harvesting head of forest harvester, the range data of the target trees were collected with a laser scanner. The backgrounds of the scanned data were filtered with the filtering algorithm based on principles of corrosion and clustering. Then, the outline of the scanned target was drawn. On the assumption that all cross sections of the target trees were standard circulars, then the radiuses of them were calculated with the least square method and the mean error of which was less than 4.29 mm. At last, a kind of clustering method based on multivariate statistical analysis was used to classify the calculated results, so the target trees and the large obstacles could be classified based on hierarchical cluster. The final experimental result showed that the method could distinguish the target trees with radius smaller than 384 mm and the large obstacles with calculated radius larger than 774 mm effectively.