Classification of Fused Imagery Base on the SFIM Algorithm

This paper aims to further study the classification accuracy of the SFIM-fused imagery based on a Landsat 7 ETM⁺ sub-scene covering the urban fringe of southeastern Fuzhou City. The study reveals that the classification accuracy of the SFIM-fused image is higher than that of the original image. Nevertheless, the difference in smoothing filter kernel sizes used in producing the SFIM-fused images can affect the classification accuracy. Too large smoothing filter kernel size will decrease classification accuracy as well as spectral fidelity in spite of increasing spatial frequency information absorption. Using a 5×5 smoothing filter has achieved the highest classification accuracy in this case study.