A Content Separation Image Fusion Approach: Toward Conformity Between Spectral and Spatial Information

For many remote-sensing applications, it is desirable to have the best possible spatial resolution in order to resolve fine features on the earth's surface. At the same time, a high spectral resolution is needed to distinguish among different ground covers. This paper presents an approach that is based on separating the spatial and spectral characteristics of features extracted from the high-frequency undecimated wavelet transform components. The extracted content, which is expressed in terms of wavelet transform modulus maxima, is used to construct the wavelet coefficients, which reflect the fine features of the more highly resolved scene and are in agreement with spectral properties of the lower resolved multispectral bands. The method effectively sharpens multispectral data and preserves the spectral characteristics of features to a great extent. The experiments have shown that the proposed framework can be applied to fuse data sets of both high- and low-correlated optical as well as near-infrared imagery of low m/p spatial resolution ratio, such as SPOT4