基于α-shape与SSA-XGBoost算法的无人机点云孔洞修补  

The repair of drone point cloud potholes based on theα-shape and SSA-XGBoost algorithms

作　　者：宋晓辉 吕富强 窦彩英 唐诗华 党梦鑫 SONG Xiaohui;LÜ Fuqiang;DOU Caiying;TANG Shihua;DANG Mengxin(College of Geomatics,Xi’an University of Science and Technology,Xi’an 710054,China;Yulin Municipal Bureau of Natural Resources,Yulin 537000,China;College of Geomatics and Geoinformation,Guilin University of Technology,Guilin 541004,China;The First Topographic Survey Team of the Ministry of Natural Resources,Xi’an 710054,China)

机构地区：[1]西安科技大学测绘科学与技术学院,陕西西安710054 [2]玉林市自然资源局,广西玉林537000 [3]桂林理工大学测绘地理信息学院,广西桂林541004 [4]自然资源部第一地形测量队,陕西西安710054

出　　处：《海洋测绘》2024年第4期69-73,共5页Hydrographic Surveying and Charting

基　　金：国家自然科学基金项目(42064003);广西自然科学基金项目(2022GXNSFBA035639);广西高校中青年教师科研基础能力提升项目(2023KY1199)。

摘　　要：针对极限梯度提升树算法在进行无人机点云孔洞修补时核心超参数难以选取、点云孔洞修补范围识别困难以及孔洞修补精度较低等问题,提出基于麻雀搜索算法优化极限梯度提升树的点云孔洞修补方法。首先利用α-shape算法进行点云的孔洞识别,在此基础上,获取点云孔洞和周围点云的位置信息并将其作为模型的输入样本。再利用麻雀搜索算法优化极限梯度提升树算法中的核心超参数,构建SSA-XGBoost点云孔洞修补模型,并将该模型应用于无人机点云孔洞的修补中。最后将SSA-XGBoost与XGBoost、BP神经网络两组算法进行预测精度的对比。实验结果表明:SSA-XGBoost模型的预测结果相较于其它两组对比算法预测精度更高,在点云孔洞修补方面具有一定的意义。Aiming at the problems of difficult selection of core hyperparameters,difficult identification of point cloud hole repair range and low hole repair accuracy when using Extreme Gradient Boosting algorithm for UAV point cloud hole repair,this paper proposes a point cloud hole repair method based on Sparrow Search Algorithm to optimize the limit gradient lifting tree.Firstly,theα-shape algorithm is used to identify the holes in the point cloud.The position information of the holes in the point cloud and the surrounding point cloud is obtained and used as the input sample of the model.Using Sparrow search algorithm to optimize the core hyperparameters in the limit gradient lifting tree algorithm,the SSA-XGBoost point cloud hole repair model is constructed,and the model is applied to the repair of UAV point cloud holes.Finally,the prediction accuracy of SSA-XGBoost is compared with XGBoost and BP neural network.The experimental results show that the prediction of SSA-XGBoost model is more accurate than the other two algorithms,which has certain significance in point cloud hole repair.

关 键 词：摄影测量 孔洞修补 α-shape算法 麻雀搜索算法 极限梯度提升树 

分 类 号：P231[天文地球—摄影测量与遥感]