【摘要】 森林蓄积量能够评估林地生产力的高低及经营措施的效果,为森林经营与采伐提供重要依据。目前,大多基于无人机影像的蓄积量估算,均建立在测绘标准所生成的DOM、DSM、DEM等测绘成果基础上,而未充分利用原始影像数据上的林业特征,无法从点云层面上加入林业业务逻辑产生成果数据。获取无人机影像后,利用特征点提取与匹配方法自动相对定向,结合控制点和光束法平差的迭代求解,解算出精确的相机姿态数据,并沿核线方向一维搜索特征点进行影像密集匹配生成密集点云。对原始三维点云过滤后进行树冠分割,在聚类后的林冠点云中提取了树顶点和树高因子估测了森林蓄积量。研究结果表明,冠幅的提取精度85.15%,树高的提取精度83.69%,林分蓄积量估算的精度达到了82.46%。更多还原

【Abstract】 The amount of forest stocks can assess the productivity of forest land and the effects of management measures, so as to provide important basis for forest management and harvesting. At present, most of the forest volume estimation based on UAV images was based on DOM, DSM, DEM and other surveying and mapping results generated by surveying and mapping standards. However, the forestry characteristics of the original image data were not suf?ciently utilized, and the forestry business logic can not be added to the point cloud level platform to produce the achievement data. After obtaining the uav image, based on the method of feature point extraction and matching and automatic relative orientation, combining the iterative solution of control point and beam adjustment, the accurate attitude data of the camera was calculated, and the feature points were searched along the direction of the core line in one dimension for dense image matching and formation of dense point cloud. After ?ltering the original three-dimensional point cloud, the canopy segmentation was carried out, and the tree vertex and tree height factors were extracted from the clustering point cloud to estimate the forest accumulation. The results indicate that the extractive accuracy of the crown breadth was 85.15%, the extractive accuracy of the tree height was 83.69%, and the estimation accuracy of the forest stand volume reached 82.46%.更多还原