Detection and diagnosis of rail surface defects using signal and image processing techniques
Regular inspection of railway tracks is essential for safe transportation of people, animals, and goods. However, the inspections are usually conducted by humans, which can be subjective, labour-intensive, time-consuming, and weather-dependent. Hence, many indirect measuring instruments and data processing methods are developed to replace human inspection, which increase reliability and efficiency. This paper (1) presents the causes and characteristics of common rail surface defects (RSDs) namely corrugation, squats, shells, spalls, and flakes, as well as (2) evaluates various signal and image processing techniques for detecting and diagnosing these RSDs. The processing techniques involve filtering, data analysis, feature extraction, selection, and classification. Simulation and field experiment results are also discussed. Furthermore, current challenges and potential future directions for research and development in this area are outlined.