Characterizing spatial structure of tree canopy using colour photographs and mathematical morphology

Abstract

In this paper, some basic principles of morphological transformation are reviewed. Algorithms ofgranulometric analysis have been developed and applied to characterize three kinds of young trees — fir, spruce, and pine. Conventional colour photographs have been used. A number of preprocessing techniques have been used to separate the tree crown from the background in each photograph, resulting in three binary tree-crown images for subsequent morphological processing. Two types of structuring elements, linear and disk-shaped, have been used to generate structure signatures such as pattern spectra through granulometric analyses of the binary images of tree crowns. Experiment results indicate that pattern spectra of size distribution parameters extracted from each binary image are considerably different among the three tree species. Crown structural pattern spectra may therefore be useful alternatives for the recognition of tree species. © Canadian Journal of Remote Sensing/Journal canadien de télédétection.