Research on Train Motor Image Localization Method Based on Infrared Feature Screening and Matching

To improve the recognition of traction motor status during train operation, this paper proposes an image localization method based on infrared feature screening and matching. Thermal imaging videos of the train undercarriage are first captured using an infrared thermal imager. Then, efficient screening and precise localization of motor images are achieved through multi-scale feature analysis, sliding window grouping, and template matching algorithms. The proposed method was validated on Nanjing Metro Line 1. Experimental results demonstrate that the method exhibits excellent robustness, real-time performance, and accuracy under varying temperature conditions. It significantly outperforms traditional approaches in terms of F1 score and frame processing efficiency, demonstrating strong engineering applicability.