Study on methods of fusion and classification using SPOT5 image of ZhongShan cemetery

In this paper, efforts were made to merge SPOT5 panchromatic image with multispectral images using six different data fusion algorithms, which were Hue-Intensity-Saturation(HIS), Principle Component Analysis(PCA), Kauth-Thomax (K-T) transform, Linear-weighted transform, Brovey fusion and Wavelet transform fusion. We evaluated the fusion images in both subjective and objective factors. The research showed that the fused images had higher spatial resolution while maintaining the basic spectral contents of the original multispectral images, the visual effects and the accuracy of the classification using fused images were improved greatly. Among the six fusion algorithms, the images using Brovey fusion, Wavelet transform and PCA transform were better, which can be applied in forest resource survey. Four classification methods were introduced to this paper, and the study region was classified into three forest types and other coverage type, which included eleven classes. Through attempting to introduce into decision tree idea, the highest classification accuracy reached to 74.60%, the whole kappa coefficient was 0.6972. It can be seen that the whole classification accuracy and kappa coefficient were improved.