面向基建验收的机载激光电力线路识别

基于激光点云的电力线路自动识别实现基建验收的全局化,已经成为机载激光雷达电力线路管理和数据处理的一个重要课题。提出了一个全新的、完全自动化和多功能的电力线路识别技术框架。该框架通过利用多尺度邻域对电力线路的空间拓扑进行表征,并在特征提取的基础上采用支持向量机(S VM)进行数据点分类,最终实现了电力线路走廊数据点云的高效率识别。通过试验评估表明,基于空间拓扑特征的数据点云自动识别技术能够改善电力线路的分类和识别的效率。

Airborne Laser Power Line Recognition for Infrastructure Acceptance

The automatic identification of power lines based on laser point clouds to achieve the overall acceptance of infra structure inspection has become an important issue for airborne laser radar power line management and data processing.A new,ful ly automated and versatile power line identification technology framework is proposed.The framework consists of four steps:filter se lection of power line candidates,n eighborhood selection,feature extraction based on spatial topology,and SVM classification.In an experimental evaluation involving 7 neighborhood definitions,26 geometric features and two data sets,it is proved that the automat ic identification technology based on spatial topological features can improve the efficiency of power line classification and recogni tion.