A Case Study of Model-Based Satellite Image Fusion

In this paper, a comparative study was conducted on image fusion models based on moderate resolution images of MODIS and TM data. Non-supervised classification were used upon original image and two fusion images (PCA and HIS) respectively. Land change area and change rate were calculated and impact of color distortion from different data fusion models was presented. The result showed: (1) only one sensor data from MODIS or TM does not work well in extracting land use information because of the limitation of spatial resolution or spectral information. (2) effects from HIS and PCA models were different. PCA image had abundance information but more distortion. Spectrum statistics showed that compared with original image mean value decreased by 13.6% and increased by 8.3% for HIS and PCA, respectively. (3) HIS model could provide more details and better spectral signatures compared with the original image. It proved to be an excellent tool for distinguishing the small differences on the image.