Phase-based motion estimation enhanced with YOLO for structural dynamic displacement measurement
摘要
This paper proposes a novel YOLO–PME framework that combines YOLO object detection with phase-based motion estimation to achieve high-precision displacement measurement of moving targets in complex backgrounds. The framework leverages YOLO to identify moving targets. It uses 2D Gabor filtering to extract phase information from the target region, reducing nonlinear phase errors caused by additional textures in complex backgrounds. This method was applied to numerical simulations, vibration table experiments, and industrial chimney oscillation monitoring in practical engineering. The results demonstrate that YOLO–PME achieves high-precision dynamic displacement measurement under complex background conditions, with strong environmental adaptability, error control capabilities, and operational efficiency. This study provides a reliable solution for the field of structural dynamic displacement measurement and has broad prospects for engineering applications.