Improving Hyperspectral Image Consistency for Vegetation Classification over the Port Hedland Coastal Region

Abstract

The brightness consistency of hyperspectral airborne data over the Port Hedland coastal region was improved through a newly developed mathematical-based technique termed normalisation. Classification of the normalised data resulted in improved spatially coherent vegetation structures, with particular emphasis on mangroves. In addition, spatial statistical analysis ensured the structures were well-defined to a suitable probability, which compared favourably to the results of an earlier commercial survey based on photointerpretation and field work.