Journal of civil and transportation engineering., Journal Year: 2024, Volume and Issue: 1(4), P. 72 - 77
Published: Dec. 1, 2024
Foreign object intrusion detection is one of the important means to ensure safe operation rail transit. The distance measurement intruding foreign conducive reminding train crew take corresponding measures in time. This paper proposes a algorithm for objects based on monocular vision. Firstly, corner and threshold segmentation methods are used obtain image coordinates black-and-white checkerboard. Secondly, regression analysis employed establish conversion model between world coordinates. Finally, an YOLOv10 network recognition coordinate system. information detected. Experimental validation shows that has error range -5 +6 cm X-axis direction -9 +16 Y-axis direction, demonstrating high accuracy.
Language: Английский