Computer vision for road imaging and pothole detection:a state-of-the-art review of systems and algorithms

作     者：Nachuan Ma Jiahe Fan Wenshuo Wang JinWu Yu Jiang Lihua Xie Rui Fan 

作者机构：Department of Control Science and EngineeringFrontiers Science Center for Intelligent Autonomous Systemsand State Key Laboratory of Intelligent Autonomous SystemsTongji UniversityShanghai 201804P.R.China Department of Civil EngineeringMcGill UniversityMontréalQC H3A 0C3Canada Department of Electronics and Computer Engineeringthe Hong Kong University of Science and TechnologyHong Kong SAR 999077P.R.China CTO OfficeClearMotion Inc.BillericaMA 01821USA School of Electrical and Electronic EngineeringNanyang Technological University50 Nanyang Avenue639798Singapore 

出 版 物：《Transportation Safety and Environment》 (交通安全与环境（英文）)

年 卷 期：2022年第4卷第4期

页      面：3-18页

核心收录：

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：the National Key R&D Program of China(Grant No.2020AAA0108100) the Fundamental Research Funds for the Central Universities(Grant Nos.22120220184,22120220214 and 2022-5-YB-08) the Shanghai Municipal Science and Technology Major Project(Grant No.2021SHZDZX0100) 

主　　题：Computer vision road imaging pothole detection deep learning image processing point cloud modelling convolutional neural networks 

摘      要：Computer vision algorithms have been utilized for 3-D road imaging and pothole detection for over two ***,there is a lack of systematic survey articles on state-of-the-art(SoTA)computer vision techniques,especially deep learningmodels,developed to tackle these *** article first introduces the sensing systems employed for 2-D and 3-D road data acquisition,including camera(s),laser scanners and Microsoft *** then comprehensively reviews the SoTA computer vision algorithms,including(1)classical 2-D image processing,(2)3-D point cloud modelling and segmentation and(3)machine/deep learning,developed for road pothole *** article also discusses the existing challenges and future development trends of computer vision-based road pothole detection approaches:classical 2-D image processing-based and 3-D point cloud modelling and segmentation-based approaches have already become history;and convolutional neural networks(CNNs)have demonstrated compelling road pothole detection results and are promising to break the bottleneck with future advances in self/un-supervised learning for multi-modal semantic *** believe that this survey can serve as practical guidance for developing the next-generation road condition assessment systems.