Laser data based automatic recognition and maintenance of road markings from MLS system. (November 2018)

• Record Type: Journal Article
• Authors: Yang, Mengmeng; Wan, Youchuan; Liu, Xianlin; Xu, Jingzhong; Wei, Zhanying; Chen, Maolin; Sheng, Peng
• Is Part Of: Optics & laser technology. Volume 107(2018)
• Page Start: 192
• Page End: 203
• Publication Date: 2018-11
• Subjects: Detection -- Recognition -- Maintenance -- Road markings -- MLS data; Optics -- Periodicals; Lasers -- Periodicals; Electronic journals; 621.366
• DOI: 10.1016/j.optlastec.2018.05.027
• Languages: English
• ISSNs: 0030-3992

Abstract

Highlights: An automatic recognition of road markings method is proposed. NCC value can be used to identify whether road marking should undergo maintenance. Multiple data with different type road markings is used to verify. The high assessment indices are obtained. A valuable solution for transport management and road safety is provided. Abstract: Mobile LiDAR Systems (MLSs) have recently been recognized as an effective way to extract road markings. Although existing studies have achieved good accuracy (about 90%) in road marking extraction, the majority of them are based on image processing methods; only a few researchers directly use laser points, especially for the recognition and assessment of road markings. This study introduces a three-step automated method for the extraction, recognition, and maintenance of road markings based on the intensity information from the MLS data: (1) an automated mechanism to filter the ground surface in laser data, (2) an adaptive block and multi-threshold method to detect road markings, (3) an automated method to achieve the classification, recognition, and assessment of road markings. Qualitative and quantitative analyses based on experimental datasets with eight types of road markings were used to evaluate the feasibility and robustness of the proposed method. Experimental results show that the average values of completeness (CPT), correctness (CRT), and F-measure of the road marking detection results are 94.35%, 98.35%, and 95.7% and the average values of CPT, CRT, and quality (QUA) of recognition results are 99.0%, 93.2%, and 92.3%, respectively, indicating that the proposed method is feasible and effective. The detection and recognition results were used to reconstruct, improve, and update the road features database; provide guidelines for road applications and maximum assistance for road maintenance; and deliver a valuable solution for maintenance and management of constructed facilities.