Fusion of multispectral and panchromatic images using new methods based on wavelet transforms - evaluation of crop classification accuracy.

One of the aims of our research team is to develop algorithms to assign automatically a crop to a cadastral parcel, matching raster information (multispectral classified images) and vectorial information (polygons defining parcel borderlines). The use of multispectral images with high spatial resolution would assist these assignations. In this work, new image-fusion methods are presented and described. These methods, based on the use of the discrete wavelet transform (DWT), are improved alternatives of the standard IntensityHue-Saturation (IHS) or Principal Component Analysis (PCA) mergers. Quantitative indicators have been used to assess the spectral and spatial quality of the images resulting when IHS PCA standard mergers and IHS PCA improved mergers are used to fuse SPOT images. Finally, the utility of these merged images for obtaining “crop distribution maps” via a supervised classification has also been tested. We used ground data to classify the original and the merged images, and also to analyze the accuracy of the resulting classified images.