基于车载LiDAR点云的杆状地物分类研究

On the classification of rod-shaped object based on point cloud of mobile LiDAR

杆状地物作为常见的公共设施,对其进行自动化精确分类是一项重要工作。文中提出基于车载LiDAR点云的杆状地物自动分类方法:首先,根据杆状地物形态特点,对其进行顶部与杆部分离;再根据杆状地物点云特征提取9种特征向量,构建特征矩阵;然后采用SVM算法,对样本集特征矩阵进行训练并构建分类模型;最后,通过构建的分类模型对测试集进行分类,并与通过先验知识设置阈值的分类算法进行对比实验,实验表明,该算法无需进行多次阈值设定即可对杆状地物进行分类,并且分类精度可达94.7%,证明该算法的正确性与优越性。

As a common public facility,the automatic classification of pole objects is an important task.A method of automatic classification of rod objects based on vehicle LiDAR point cloud is proposed.First of all,according to the characteristics of the pole shape,the top and the rod are separated;nine feature vectors are extracted according to the feature of the point cloud,and the feature matrix is constructed.Then,the SVM algorithm is used to train the feature matrix of the sample set and build the classification model.Finally,the test set is classified by constructing the classification model,compared with the classification algorithm based on prior knowledge.Experiments show that the algorithm does not need multiple thresholds to classify the rod objects,of which the classification accuracy can reach 94.7%.The correctness and superiority of the algorithm is verified.