Accuracy Verification of Structural Displacement Monitoring Based on Machine Vision Measurement

In structural health monitoring, displacement is one of the key parameters reflecting the structural safety status. Traditional contact-based displacement monitoring methods suffer from problems such as complex sensor deployment, difficulty in multi-point synchronization, and insufficient real-time performance in large structures. Machine vision, as a non-contact measurement technology, provides a new technical approach for structural displacement monitoring. To verify the actual measurement accuracy of a certain machine vision two-dimensional displacement measurement system, this paper constructs an indoor test platform and sets up multi-condition experiments including static positioning and dynamic tracking. The system accuracy is tested and evaluated within a distance range of 3.5 m to 5.7 m, and error analysis is conducted using the measurement results of a sub-millimeter displacement calibration device and an electronic universal testing machine as benchmarks. The results show that within a displacement range of 100 mm, the system can stably achieve target displacement measurement and tracking. The overall error of single-target measurement is smaller than that of multi-target measurement, with a maximum error of 1.11 mm, indicating that the vision measurement system has stable measurement performance within a certain distance range.