基于面向对象的露天花岗岩矿山信息提取技术研究

Study on object-oriented information extraction technology in open-pit granite mine

矿山环境提取一直是遥感技术在矿山调查监测和治理应用中的重要环节。采用国产高分二号遥感数据,基于面向对象的分类方法,对山东省招远市北部花岗岩矿区独特的光谱特征和纹理特征进行分析,找出合适的特征参数并确定其阈值,从而构建决策树算法实现了研究区露天花岗岩矿山边界的自动提取,提取总体精度达到86.29%,Kappa系数达到0.807。

The extraction of mine environment has always been an important part of remote sensing technology in mine monitoring and treatment applications. Based on object-oriented classification method,this study used domestic high-resolution GF-2 remote sensing data to analyze the unique spectral features and texture features of the northern granite mining area in Zhaoyuan City,Shandong Province. Appropriate feature parameters were chosen and the threshold was determined. The decision tree algorithm was constructed to realize the automatic extraction of the open-pit granite mine boundary in the study area. The overall accuracy of the extraction was 86.29%,and the Kappa coefficient reached 0.807.