(北京交通大学 轨道交通控制与安全国家重点实验室，北京 100044)
(State Key Laboratory of Rail Traffic Control and Safety,Beijing Jiaotong University,Beijing 100044,China)
Urban rail transit is mainly composed of pantograph-catenary systems, track lines, vehicles, stations, etc. traditional methods such as manual inspections have poor detection efficiency, high labor intensity, and low automation and intelligence, which have brought huge challenges to the operation assurance and further healthy development of urban rail transit. At present, due to the in-depth development of vision technology, as an important detection method, machine vision has been widely used in the field of urban rail transit system state detection. In view of this, the researches and applications of machine vision in the safety state inspection of urban rail transit systems are reviewed. Firstly, the basic concept of urban rail transit and the challenges and opportunities faced by rapid development are briefly introduced. Then, the researches and applications of the machine vision technology in the safety state detection of urban rail transit subsystems are discussed in detail: 1) aiming at the problem of pantograph-catenary system state detection, the domestic and foreign research status of machine vision in pantograph wear detection, pantograph envelope line and other defects detection, catenary geometric parameter detection, catenary wear detection and catenary suspension defect detection are introduced, respectively. 2) As for track line safety state inspection, the application and research status of machine vision in fastener safety state detection and rail surface disease detection are introduced. 3) The application and research progress of machine vision in vehicle state detection are introduced in detail from the perspective of different detection items. 4) The specific application and research of machine vision in abnormal behavior detection of station escalator safety monitoring and platform safety monitoring are sorted out and summarized. 5) The specific application and background technology of machine vision in rail transit driver behavior monitoring are presented. Finally, we look forward to the future of the application of machine vision technology in the field of urban rail transit system state inspection.