Detecting road poles from mobile terrestrial laser scanning data

Abstract

Detecting road poles from mobile terrestrial laser scanner (MTLS) point clouds is important for many geographical information system (GIS) applications such as right-of-way asset inventory compilation. The aim of this research is to automatically detect the road poles from unorganized 3D point clouds captured by an MTLS system named TITAN. The proposed pole detection pipeline consists of a sequence of five steps: organizing the 3D point clouds and nearest neighbor search, 2D density-based segmentation, vertical region growing, segment merging, and pole classification. The obtained average detection rate and accuracy for the three data sets tested were 86% and 97%, respectively.