基于植被指数的GF-2影像防护林快速提取研究

Research on Rapid Extraction of Shelter Forest Using GF-2 Images Based on Vegetation Indices

以0.8 m的GF-2卫星影像为主要数据源,根据防护林特有植被指数特征,研究GF-2遥感影像通过植被指数快速提取防护林的方法。通过样本点特征分析和对比分析分别确定比值植被指数(RVI)、差值植被指数(DVI)和归一化差值植被指数(NDVI)的提取阈值,辅助Landsat8 OLI影像做掩膜去除建设用地中林地的干扰信息,对影像做二值化处理,分别得到3种植被指数防护林的提取结果。研究结果表明,用RVI、DVI以及NDVI分类的林带识别总体精度分别为89.2%、90.4%和88.3%,均在85%以上,说明这三种方法都可以较精确地提取防护林信息,其中采用DVI的方法提取效果相对较好,同时在林带交错及林带间距较小时,DVI可提供更为丰富的细节信息,有助于林带信息的精准识别。本研究可在大尺度下为东北地区防护林的现状监测及评价提供科学支撑。

In this study, 0.8 m GF-2 satellite image was taken as the main data source. According to the vegetation index characteristics of shelter forest, the method of GF-2 remote sensing image to quickly extract shelter forest through vegetation index was studied. The extraction thresholds of the ratio vegetation index （RVI）, the difference vegetation index （DVI） and the normalized difference vegetation index （NDVI） were determined by sample point feature analysis and comparative analysis, respectively. The Landsat80LI image was used as a mask to remove the interference information of the forest land in the construction land, and the image was binarized to obtain the extraction results of the three vegetation indexes of the shelter forest. The results show that the overall accuracy of forest belt identification using REi, DEIand NDEI are 89.2%, 90.4% and 88.3%, respectively, both are above 85%, indicating that these three methods can extract the information of shelter forest very accurately. Among them, DEI method has relatively accurate extraction result, and it can provide more detailed information when the forest belt is interlaced and the forest belt spacing is small, which is helpful for the accurate identification of forest belt information. This study can provide scientific support for the monitoring of current status and remote sensing inversion of large-scale sheherbehs in Northeast China.