车载点云特征对应影像自动分割技术研究  

Research on automatic image segmentation based on vehicle-mounted point cloud feature

作　　者：钟少忠 周克军 王留召[2,3,4] 毛明楷 ZHONG Shaozhong;ZHOU Kejun;WANG Liuzhao;MAO Mingkai(Guangdong Surveying and Mapping Product Quality Supervision and Inspection Center,Guangzhou 510075,China;Capital Normal University,Beijing 100089,China;Chinese Academy of Surveying&Mapping,Beijing 100036,China;Beijing Geo-Vision Information Technology Co.,Ltd.,Beijing 100070,China)

机构地区：[1]广东省测绘质量监督检验中心,广东广州510075 [2]首都师范大学,北京100089 [3]中国测绘科学研究院,北京100036 [4]北京四维远见信息技术有限公司,北京100070

出　　处：《测绘通报》2022年第6期45-48,97,共5页Bulletin of Surveying and Mapping

基　　金：国家自然科学基金(42071444);广东省自然资源厅基础测绘研究项目。

摘　　要：车载点云数据通过自动提取,可快速得到大量的三维特征数据,利用同步影像数据不仅能提高目标判读的准确性与效率,而且能获取目标对象的属性信息,对质检巡视、部件属性调查、三维建模等生产活动有重要的作用。本文提出了一种车载点云特征数据对应的影像自动分割技术,通过影像的位置姿态判断影像与特征信息间的关系,实现最佳影像自动选取和分割,生成适当范围局部正射影像,并在新影像中标示出特征信息的正确位置。Through automatic extraction of on-board point cloud data,a large number of three-dimensional characteristic data can be quickly obtained.Using synchronous image data can not only improve the accuracy and efficiency of target interpretation,but also obtain the attribute information of the target object,which plays a very important role in production activities such as quality inspection,component attribute investigation and 3 D modeling.In this paper,an automatic image segmentation technology corresponding to the feature data of vehicle-mounted point cloud is proposed.The relationship between the image and the feature information is judged by the position and attitude of the image,so as to realize the optimal image automatic selection and segmentation,generate the appropriate local orthographic image,and mark the correct position of the feature information in the new image.

关 键 词：车载移动测量系统 点云 三维特征数据 影像数据 影像自动分割 

分 类 号：U463.6[机械工程—车辆工程] TP391.41[交通运输工程—载运工具运用工程]