资源一号02D卫星高光谱与多光谱融合数据滨海湿地分类应用

Classification of Coastal Wetlands Based on Hyperspectral and Multispectral Fusion Data of ZY-1-02D Satellite

针对当前影像融合方法虽然取得了很大进步,但仍然存在光谱与空间失真的问题,提出了一种适用于资源一号02D卫星滨海湿地分类应用的影像融合方法,可实现较好的影像融合质量,提升资源一号02D卫星数据的滨海湿地分类精度。通过计算数据波段相关性对高光谱影像进行波段分组,并采用成分替换方法与神经网络相结合实现资源一号02D卫星数据的空间分辨率提升。在模拟数据集总定量指标获得了最佳值,在真实数据集中融合影像光谱曲线与参考光谱曲线具有较小差异,分类应用中总体分类精度达到0.97,Kappa系数为0.93,结果表明:提出的方法能够很好地提升资源一号02D卫星的影像融合质量,可进一步提升卫星高光谱影像在滨海湿地精细分类等方面的应用价值。

Although current image fusion methods have made great progress,there are still spectral and spatial distortion problems.An image fusion method suitable for ZY-1-02D satellite coastal wetland classification application is proposed.The applied image fusion method can achieve better image fusion quality and improve the classification accuracy in coastal wetland based on ZY-1-02D satellite data.Hyperspectral images are grouped by calculating band correlations and a component replacement method is combined with neural network to achieve spatial resolution improvement of ZY-1-02D satellite data.In simulation datasets,the total quantitative index has obtained the best value and the fusion image spectrum in the real dataset has a small difference with the reference spectrum.In the classification application,the overall classification accuracy reaches 0.97 and the Kappa coefficient is 0.93,indicating that the method proposed in this paper can improve the fusion quality of ZY-1-02D satellite,which can further enhance the application value of the hyperspectral data of ZY-1-02D satellite in fine classification of coastal wetlands.