Camille Prost, Paul Dare, Andre Zerger
Department of Geomatics, The University of Melbourne
Victoria 3010, Australia
Tel: +61 3 8344 9901; Fax: +61 3 9347 2916
Email: [email protected], [email protected], [email protected]
Abstract
Eucalypt tree dieback is a disease that threatens the survival of native forests in Victoria (Australia). With the aim of mapping and monitoring the spatial distribution of dieback, airborne imaging technologies can be more effective than ground surveys. Amongst the numerous types of airborne sensors, the linescanner and the video camera provide images with high spatial resolutions. In order to detect individual defoliated Manna Gum trees at Mt Eccles National Park (south-western Victoria), both linescanner data consisting of several transects across the study site, and high spatial resolution airborne video were captured. However, highlighting the health status of sparse and mainly unclustered defoliated eucalypts requires some complex algorithmic processing of the data. Thus far, a canopy delineation method based on supervised Markovian texture modelling was developed. To account for the random spatial nature of eucalypt canopies, the texture corresponding to the canopy area was modelled using a parametric Markov random field. The procedure enables better vegetation health classification by principal component analysis. Additionally, pre-processing of the video frames was performed to remove the motion blur present in the images. Defoliated trees are then tracked down by a pattern recognition routine using a neural network trained to locate classes of eucalypts with different leaf densities. Both the processed data sets yield useful information for highlighting vegetation dieback. Thus, the two approaches presented in this paper provide complementary means of assessing eucalypt tree health and consequently assist park managers with managing dieback.
