Fusion of remotely sensed images based on subsampled contourlet transform and spectral response

摘要

The objective of the remote sensing image fusion involves combining the desirable characteristics of a low-spatial resolution multispectral (MS) image and a high-spatial panchromatic (PAN) image to obtain a single multispectral image with high spatial resolution. The traditional image fusion approaches, including component substitution, detail injection, discrete wavelet transform (DWT), high-pass filtering technique and so on, often cause the loss of spectral information or degrade the spatial quality. Although the DWT-based image fusion algorithms are popular in the image fusion field, the computational load of the undecimated version of DWT is comparatively high and the artifacts of merged images with the decimated DWT are apparent. A novel image fusion algorithm based on the subsampled contourlet transform and spectral response is proposed in this paper. The proposed algorithm outperforms the conventional image fusion techniques with higher spatial quality and better visual effects.