Context-driven image fusion of multispectral and panchromatic data based on a redundant wavelet representation

A method of image data fusion is proposed, which applies a stationary or redundant wavelet decomposition to merge multispectral and panchromatic images. Couples of subbands of corresponding frequency content are locally merged if a correlation condition is satisfied, and the fused image is produced by taking the inverse transform. A local correlation coefficient between the low-frequency subband of the high-resolution image and the corresponding subband of the up-sampled multispectral image is used in order to pilot the fusion process. The experimental results prove that the method incorporates contextual spatial information and reduces possible over-enhancement of the multispectral data on the image regions having higher spectral content, as in the case of urban areas or other highly textured regions.