基于高光谱SMPI法的草原地表微斑块识别与分类

Identification and classification of surface micrograss on grassland based on Hyperspectral SMPI method

草原地表微斑块特指在高光谱图像上荒漠化草原鼠洞、裸土和植被所表现出的斑块。这些是评价草原退化的重要指标。地表微斑块的识别与分类是基于遥感草原退化监测与评价研究的基础性的工作,对草原区域性规划、系统恢复与重建理起到至关重要的作用。由于高光谱采集及分析诸多方面的原因,目前地表微斑块高光谱的识别研究较少。本研究首次以荒漠化草原实地地表微斑块为研究对象采集高光谱数据,提出了SMPI法(Surface Mini-Patch Index,地表微斑块指数法),实现了草原地表微斑块的高精度识别与分类。研究结果表明,SMPI法可将荒漠化草原地表微斑块进行高精度识别与分类,通过Kappa系数验证,整体验证精度高于96%。此研究为无人机或低空遥感进行草原退化监测与定量反演提供了基础。

The micrograss patch on the grassland surface refers specifically to the patches shown by the mouse holes, bare soil, and vegetation on desertified grasslands in hyperspectral images. These are impor- tant indicators for evaluating grassland degradation. The identification and classification of surface micro- plaques is based on the basic work of remote sensing grassland degradation monitoring and evaluation re- search, and plays a crucial role in grassland regional planning, system restoration and reconstruction. Due to the reasons of hyperspectral acquisition and analysis,there are few studies on the identification of hy- perspectral microscopic plaques. This study collected the hyperspectral data from the surface micro- plaques in desert grasslands and proposed the surface mini-patch index （SMPI） method to achieve high- precision identification of grassland surface micro-plaques classification. The research results show that the SMPI method can identify and classify the surface micro-plaques in desertification grasslands with high accuracy. The Kappa coefficient verification shows that the overall verification accuracy is higher than 96 %. This study provides the basis for monitoring and quantitative inversion of grassland degrada- tion using drones or low-altitude remote sensing.