GF-2卫星数据融合及在矿山地物识别中的应用

GF-2 Satellite Data Fusion and Its Application in Mine Surface Feature Recognition

文章以GF-2国产卫星数据为研究对象,以某矿山区为试验区,就多光谱与全色波段数据融合图像的矿山地物目视解译为目的,从GS变换法、Brovey变换法、PCA变换法和NNDiffuse算法开展融合试验研究,对4种方法融合的图像从空间细节的加强和光谱信息的维持2个方位进行主观视觉和客观定量进行了比拟评判。结果表明:NNDiffuse算法融合的图像在亮度信息、信息量、清晰度和光谱信息4个方面具有综合优势;此外,应用研究表明:基于NNDiffuse算法融合的图像对于大于9 m 2的不同矿山地物目标均能够识别和圈定。

Taking GF-2 domestic satellite data as the research object and a mine area as the test area,this paper studies the integration experiments of the GS transformation method,Brovey transformation method,PCA transformation method and NNDiffuse algorithm for the visual interpretation of mine surface features of multi-spectral and panchromatic band data fusion images.The images fused by the four methods are compared and evaluated by a subjective vision and objective quantification from the enhancement of spatial details and the maintenance of spectral information.The results show that the images fused by the NNDiffuse algorithm have comprehensive advantages in brightness,information content,definition and spectral information.In addition,application research shows that the images fused by the NNDiffuse algorithm can identify and delineate different mine surface features larger than 9 m 2.