Geometric and radiometric constraints-based extraction of urban road manhole covers and their maintenancerelated information using mobile laser scanning data.

In recent years, mobile laser scanning (MLS) has attracted increasing interest for road network three-dimensional mapping, where one of the applications of MLS data is automation in urban road manhole covers detection and their information measurement for monitoring and management of manhole covers on regular-basis. Therefore, a multi-step method is proposed based on geometric and radiometric constraints on manhole covers for their extraction and maintenance-related information measurement. The proposed method was implemented using graphics processing unit (GPU)- based parallel computing framework with improved computation time. The method performance was evaluated using MLS datasets of two complex urban roadway test sites, which included partially overlapping objects, occluded road surfaces, and intra-class variability. The manhole covers were extracted at average completeness, correctness and quality of 96.69%, 98.32%, and 95.12%, respectively. The proposed method performed satisfactorily in the challenging cases of road manhole covers and achieves significant improvement in comparison with several state-of-the-art methods. [ABSTRACT FROM AUTHOR]