期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2007
卷号：XXXVI-3/W49A
页码：129-134
出版社：Copernicus Publications
摘要：A method to automatically detect the tree crowns shape is presented in this paper. The study site is located in some mountainous parts of Friuli Venezia Giulia characterized by coniferous, mixed and broad-leaved forests with different population densities. The method, developed in an open source environment, is based on mathematical morphology operations that assess the cartographical position of the trees, as well as the height of the trees. Starting from single-extracted trees, a segmentation algorithm makes it possible to classify the laser point data as a subset of crown points. Then, the crowns are delineated by circular polygons centred on the geometric laser point barycentre. To enhance the quality of the calculated crown parameters (area, depth of insertion, volume), a statistical analysis of the height (z) frequency distribution was performed which allows the re-filtration of the low vegetation (border or under-canopy vegetation). The results have been validated using topographic total station data surveyed in situ, in 13 forestry sample plots with a total of about 550 reference trees. Considering the ecological diversity (complexity) of the chosen plots, the paper shows a high correlation between plot data and laser scanning extracted data, particularly in the coniferous areas, underlining the possibility of extending the fields of research to the study of the dominated vegetation under canopy
关键词：LiDAR; Tree extraction; Morphological analysis; Region Growing; Crown delineation; Forest typology
