Automatic identification method of foreign body intrusion in railway transportation track based on improved LeaderRank identification of key points

Abstract

To accurately identify foreign object intrusion behaviors in key areas of railway transport tracks, an automatic recognition method is proposed. This method is based on an improved LeaderRank algorithm and is designed to detect foreign object intrusions on railway tracks. First, the improved LeaderRank algorithm identifies key points of trajectories, which are then used as layout points for installing monitoring equipment. Real-time monitoring devices collect video images of key track areas. Next, an improved Gaussian mixture model is used for image segmentation in track monitoring, extracting potential foreground images containing foreign object intrusions. These images are then input into a hybrid deep learning-based automatic recognition model for foreign object intrusion. The firefly algorithm trains this model, constructing a structurally stable hybrid deep learning model that learns the relationship between image combination features and foreign object intrusion behaviors, enabling accurate recognition of foreign object intrusions. Experimental results demonstrate that this method accurately identifies foreign object intrusion, enhancing detection accuracy and reliability. The proposed method, combining the improved LeaderRank algorithm with hybrid deep learning, offers an efficient and accurate solution, providing a new technical approach for railway transport safety management.