A systematic literature review of defect detection in railways using machine vision-based inspection methods

Train rolling stock and track inspections are necessary for the safe operation of the train. For this reason, a regular inspection of defects is required for the train rolling stock. The conventional defect detection methods yield low efficiency, consume more time, are unreliable, and are less cost-effective. These obstacles may be mitigated by integrating a machine vision-based inspection system (MVIS). This systematic literature review explores the landscape of railway defect detection methodologies, primarily focusing on leveraging image processing techniques. This comprehensive analysis encompasses many studies examining the evolution of image processing applications in the context of railway rolling stock and rail track defect detection. From traditional methods to the latest advancements, a nuanced understanding of the challenges and innovations in this domain is required. Key themes include utilizing computer vision algorithms, machine learning models, and deep learning techniques for enhanced accuracy in identifying defects. We delve into the intricacies of image acquisition, preprocessing, and feature extraction, shedding light on the pivotal role of these processes in refining defect detection systems. Also, the current gaps and opportunities for future research, emphasizing the need for standardized datasets, benchmarking methodologies, and the integration of emerging technologies, are highlighted. This review not only consolidates the existing knowledge, but also serves as a roadmap for researchers invested in advancing the field of railway defect detection. By synthesizing insights from many studies, this review contributes to a deeper understanding of the state-of-the-art in railway defect detection using image processing, fostering dialogue and collaboration for improving railway safety and reliability.