The feasibility of extracting urban fringe land cover information from pixel-level fused TM and SPOT images

Abstract

Given a set of remotely sensed images with different spatial and spectral resolutions, information preserving techniques of pixel-level image fusion can be used to generate a synthesised image that maximises the preservation of both the spectral and the spatial information available from the source images. These techniques have the potential for improving the quality of information on urban fringe land cover patterns, which are important to many applications. Here, we assess the feasibility of extracting useful land cover information in order to derive more accurate spatial information to inform sustainable land-use of the urban fringe. After a brief review of some existing pixel-level image fusion techniques, the results from applying a wavelet transform based pixel-level image fusion technique on subsets of TM and SPOT images over a portion of the eastern fringe of Melbourne are evaluated in terms of spectral fidality and spatial enhancement. We argue that the wavelet fusion technique demonstrated here may provide an efficient and cost-effective means to support enhanced land cover and land use assessment.