Superresolution of IKONOS multispectral images using variational bayes and fast IHS transform

By incorprating spatial detail information of panchromatic image into multispecral images, the superresolution reconstruction (SRR) or pansharpening of multispectral images can be achieved. Many methods for SRR of multispectral image have been proposed, such as IHS-based SRR methods, PCA-based SRR methods, and wavelet transform based SRRs. However, further studies is still desirable for new type of remotely sensed images, such as IKONOS imagery. In this paper, we propose a SRR method for IKONOS multispectral images based on Baysian method. Firstly, the fast IHS transform (FIHS) of multispectral images of IKONOS images is performed; then intensity component I and IKONOS panchromatic image are fused into a new panchromatic image by the Baysian method. After that, with the new intensity component, the FIHS transform is performed, thus the high resolution of multispctral images is obtained. Experiment results on IKONOS mutlispectral images show the effectiveness of the proposed approach.