Design and Optimization of a Dynamic Train Image Acquisition System Utilizing Line Scan Camera Technology
Changgeng Li, Xu Feng, Wei Li, Xiaodong Sun, Bo Yuan · 2023
In response to the demand for efficient and high-quality daily maintenance and inspection of modern railway trains, the utilization of AI-based machine vision technology has been proven to significantly enhance maintenance efficiency and quality. Image quality plays a critical role in visual inspection. Thus, this study presents the design of a dynamic train exterior image acquisition system based on line scan camera technology. Through a comprehensive analysis of maintenance requirements and business scenarios, the hardware structure and functional framework of the system are meticulously developed. The image acquisition module is strategically deployed in a 360° arrangement to enable panoramic image acquisition of the train’s exterior. Leveraging the imaging principle of the line scan camera, two sets of magnetic cylinders are employed to measure the train’s velocity multiple times, ensuring continuous correction of the image acquisition line frequency for stable image capture. Furthermore, a character recognition algorithm for train numbers is extensively researched, enabling accurate identification of the train number through image acquisition, preprocessing, and OCR recognition algorithms. The proposed system integrates professional image acquisition and processing techniques, enabling high-resolution and high-precision stable acquisition of dynamic train images. It provides crucial data support for precise detection of train exterior conditions and offers reliable technical means for train trajectory monitoring and safety management.