No abstract

No abstract

Because of the trade off between spatial resolution and spectral resolution in satellite imagery, it is often desirable to fuse lower resolution multispectral imagery with a high-resolution panchromatic image in order to obtain an image with the spectral resolution and quality of the former and the spatial resolution and quality of the latter. In an urban setting, the spectral information can be used to discriminate between the numerous different land cover types, both natural (vegetation) and human generated (roads and buildings), while the spatial information can be used to clearly delineate their boundaries. Standard image fusion methods, such as methods involving IHS or PCA, are often successful at injecting spatial detail; however, they tend to distort the colour information. The potential benefits of wavelet-based image fusion methods have recently been explored in a variety of fields and for a variety of purposes, in particular for fusing panchromatic and multispectral imagery. In this paper, the results from a number of wavelet-based image fusion schemes are compared in terms of their similarities and differences, and their advantages and limitations. It was found that, while even the simplest wavelet-based fusion scheme tends to produce better results than standard fusion schemes such as IHS and PCA, particularly in terms of minimizing colour distortion, decimated and undecimated algorithms often disturb the linear continuity of spatial features. The results from wavelet-based methods can be improved by applying more sophisticated schemes or more advanced models for injecting detail information; however, these schemes are more computationally complex and often require the user to determine appropriate values for certain parameters, such as thresholds. More comprehensive testing is required in order to fully assess under what conditions each approach is most appropriate.

No abstract