Landsat 7 ETM+全色与多光谱数据融合算法的比较

On fusion algorithms of Landsat 7 ETM+ PAN and multi-spectral images

以Landsat 7 ETM+数据为例,采用Brovey变换、HSV变换、PCA变换以及Gram-Schmidt光谱锐化算法对比研究了同一传感器全色和多光谱数据的融合问题.由定性和定量分析认为:空间分辨率的增强和光谱特性的改善是相互矛盾的,针对Landsat 7 ETM+本身的融合而言,综合考虑保光谱特性、信息量和清晰度,最佳的融合算法是PCA变换,而Gram-Schmidt光谱锐化法要优于Brovey变换和HSV变换.该研究为充分利用Landsat 7 ETM+的全色高分辨率和多光谱特性,进一步挖掘Landsat 7 ETM+的数据潜力奠定了基础.

Four different fusion algorithms such as Brovey, HSV, PCA and Gram-Schmidt spectral sharpening were compared when the panchromatic and multi-spectral images of Landsat 7 ETM＋ were merged. From the qualitative and quantitative analyses we concluded that the enhancement of spatial resolution is inconsistent with the improvement of spectral characteristics. With consideration of the spectral information fidelity, information content and the improvement of spatial detail information, the experiment indicates that PCA is the best fusion algorithm, and Gram-Schmidt is better than Brovey and HSV. We believe that this research has laid a foundation for the fusion of Landsat 7 ETM＋ panchromatic and multi-spectral images.