The Extraction of Individual Tree-Crown in Aerial Digital Camera Imagery

A high spatial resolution remote sensing image provides a huge amount of data about our earth surface in more details, and makes it possible to estimate forest and tree parameters directly from aerial image. In this research the segmentation algorithm of region-merging technique was used for the identification of the individual tree crown. Fuzzy logic membership function and nearest neighbor classifier were used for the extraction of tree canopy information. The crown canopy was extracted firstly based on the multi-scale segmentation, fuzzy logic classification, and objects merge based on the class. A segmentation of tree crown mode was conducted for the individual tree crown. To get satisfying outcome, morphological reconstruction based on fuzzy logic functions, such as the feature of the area threshold and relative border to, were used to eliminate some false polygons. The results from the image-derived and field measured tree crown comparisons in this paper show that there was significant correlation, with the correlation coefficients varying from 0.69 to 0.82. Most of the tree crowns were underestimated because of the influence of shadow and crown overlap. And further research will concentrate on individual trees with little crown overlap and full rounded crowns.