基于区域生长优化的水利工程设施点云单体化研究  

Research on Point Cloud Monomer in Water Conservancy Project Facilities Based on Regional Growing Optimization

作　　者：杨夏森 张晓春[1] 项兴华 邵广俊 胡伟飞 YANG Xia-sen;ZHANG Xiao-chun;XIANG Xing-hua;SHAO Guang-jun;HU Wei-fei(State Key Laboratory of Water Resources Engineering and Management,Wuhan University,Wuhan 430072,China;State Grid Zhejiang Electric Power Co.,Ltd.,Jinshuitan Hydropower Plant,Lishui 323000,China)

机构地区：[1]武汉大学水资源工程与调度全国重点实验室,湖北武汉430072 [2]国网浙江省电力有限公司紧水滩水力发电厂,浙江丽水323000

出　　处：《水电能源科学》2024年第7期216-220,共5页Water Resources and Power

基　　金：国网浙江省电力有限公司科技项目(B311JS230001)。

摘　　要：水利工程中输变电设施安全监测是保证水电站安全运行的重要环节,其中基于点云的变形监测需要准确分割出各种输变电设施。为此,通过对三维点云进行降维处理,获取水利工程设施的二维点云图像,基于区域生长算法和点云单体化算法实现分割工作。同时在区域生长算法的基础上提出了一种旋转角度优化算法,通过旋转三维点云数据,得到易于分割的最优角度点云图像,提高了区域生长算法的精度和速度。为验证旋转角度算法的精度,设计了仿真试验分割立柱点云,试验结果精确率达98%以上,表明旋转角度算法可应用于结构点云单体化领域。实际水利工程应用结果表明,所提方法可有效解决普通像素图像过分割的问题,能准确分割水利工程不同区域及设施。The safety monitoring of power transmission and transformation facilities in water conservancy projects is an important part of ensuring the safe operation of hydropower stations.Deformation monitoring based on point cloud requires accurate identification and segmentation of various power transmission and transformation facilities.The three-dimensional point cloud data was reduced in dimension for obtaining two-dimensional point cloud images of water conservancy project facilities.The region growing algorithm and point cloud monomer algorithm were used to implement the identification and segmentation work.An rotation angle optimization algorithm was proposed based on the region growing algorithm.By rotating the three-dimensional point cloud data,an optimal angle point cloud image that is easy to segment was obtained,which improves the accuracy and speed of the region growing algorithm.To verify the accuracy of the rotation angle algorithm,a simulation experiment was designed to segment column point clouds,and the experimental results show an accuracy rate reaches over 98%.This result indicates that the rotation angle algorithm can be applied in the field of structural point cloud monomerization.In practical applications in water conservancy projects,the approach presented in this paper can effectively solve the problem of over-segmentation in ordinary pixel images and accurately segment different regions and facilities of water conservancy projects.

关 键 词：点云单体化 旋转角度算法 区域生长算法 水利工程设施分割 

分 类 号：TM564.1[电气工程—电器] P258[天文地球—测绘科学与技术]