机载激光点云数据处理方法研究
Research on Data Processing Methods for Airborne Laser Point Clouds

针对机载激光扫描点云数据中杆状物分类精度不理想的问题，本文从杆状物的空间形态特征出发，构建一种基于支持向量机(SVM)模型的杆状物分类算法。首先，根据杆状物空间形态特征确定10个特征值并建立特征矩阵；其次，进行SVM模型训练并建立分类模型；最后，使用训练好的最优SVM模型进行杆状物分类。选取某段城市道路点云数据进行试验，结果表明，本文分类模型无需人工干预与阈值设定，自动化程度高，其中杆状物的最高分类精度能够达到94.23%，验证了该算法的有效性与优越性，可为基于激光点云数据的地物分类提供一定借鉴与参考。
In response to the problem of unsatisfactory classification accuracy of rod shaped objects in airborne laser scanning point cloud data, this paper constructs a rod shaped object classification algorithm based on support vector machine (SVM) model, starting from the spatial morphology characteristics of rod shaped objects. Firstly, determine 10 characteristic values and establish a feature matrix based on the spatial morphology characteristics of the rod-shaped object; secondly, conduct SVM model training and establish a classification model; finally, use the trained optimal SVM model for rod object classification. Selecting a certain section of urban road point cloud data for experimentation, the results show that the classification model in this paper does not require manual intervention or threshold setting, and has a high degree of automation. The highest classification accuracy of pole shaped objects can reach 94.23%, verifying the effectiveness and superiority of the algorithm. This can provide certain reference and guidance for land object classification based on laser point cloud data.