Automated Object Extraction from MLS Data: A Survey

Abstract

Realistic 3D city modeling using Mobile Laser Scanning (MLS) technique experienced a remarkable revolution in aiding urban planning, regulation design, city management, navigation, and emergency responses. This paper focuses on thoroughly examining the advance of automated MLS object extraction techniques over the last five years. Categorized as either on-road or off-road, mainly six objects are included in this paper (road curbs, road markings, pavement cracks, building facades, pole-like objects and trees). MLS extraction techniques applied on typical objects is evaluated according to their method design, degree of automation, precision, and computational efficiency. Recent researches mostly focus on developing accurate object extraction algorithms and most of the reviewed methods can achieve high precision, however, optimizing the trade-off between computational cost and accuracy remains a big challenge. In addition, there is still no general standardized approach to deal with MLS objects extractions to date, most algorithms reviewed in this paper still need some artificial interference to ensure accuracy and efficiency.