Abstract: Subway has become the main means of transport in large cities, and how to protect the safety of passengers has become a hot issue in subway operations. Based on manual detection between the door and platform screen doors (PSD), this article proposed a foreign object detection algorithm based on computer vision technology. The algorithm was used to detect the integrity of vehicle tail lamp belt, judge whether there was foreign objects between the door and the PSD, and then assist the driver to carry out safety inspection before driving, ensure the safe operation. Video data were collected on the spot for experimental analysis, and the results showed that the proposed method could accurately detect the foreign objects in the gap.