MULTISPECTRAL IMAGE DATA FUSION UNDER A BAYESIAN-APPROACH

Multispectral image data fusion is understood here as a process that g enerates synthetic images from a combination of primary images, by att empting to preserve the best characteristics of each primary image. Th e obtained product is important in helping the users of remote sensing in visual analysis. This paper describes a new method for multispectr al image data fusion using a Bayesian framework. As an example, the me thod is applied to the synthesis of new bands for the SPOT satellite. The method comprises two steps: (a) a new interpolator for the multisp ectral bands, obtained through the orthogonality principle, and leadin g to the estimator and its covariance matrix, which will be used as a priori information for the second step; and (b) a new statistical synt hesis formulation, also obtained through the orthogonality principle, using as observations the panchromatic and the multispectral bands. Ex perimental results with SPOT images over Guarulhos Airport, Sao Paulo, Brazil, are presented, including both the interpolated multispectral bands and the synthetic bands. A discussion of the visual output also includes a comparison between the new interpolation process with conve ntional interpolators such as nearest neighbour and bilinear. Furtherm ore, the flexibility of the Bayesian interpolator is obtained through the possibility of using different horizontal and vertical correlation coefficients that adapt the model to the local characteristics of the image being interpolated. The proposed interpolation method also allo ws the use of a simple unsharp masking procedure, with improved visual edge delineation.