Comparison of data fusion methods with high preservation based on multi-spectral and panchromatic images

摘要

The original principle of image fusion based on pixel level is using the spatial and spectral information from different remotely sensed data to generate a new image. So, the fusion method with high preservation is the main issue. In this paper, modified Brovey transform (MBT) has been proposed based on the principle of the Brovey transform model. Three fusion methods of MBT, wavelet transform (WT) and smoothing filter-based intensity modulation (SFIM), which can merge each band images directly, are applied to respectively merge multi-spectral data with panchromatic image of ETM+ and QB sensors. The qualitative evaluation and quantitative computation analysis show that MBT has the highest high frequency information preservation and SFIM model enjoys the best low frequency information preservation, and both of them can be used to deal with large numbers of images fusion for their fast computation capability. The WT has a suboptimal spectral maintenance and the lowest high spatial frequency gain among the mentioned three data fusion algorithms, and it takes more time to finish the progress of data fusion.