Abstract:In response to the widespread demand for real-time safety monitoring of low-impact traffic pavement settlement in underground projects such as tunnels and pipelines that pass under expressways and urban roads in recent years, a non-contact intelligent monitoring technology based on machine vision is proposed. For this technology, a variable-focus dome camera (optical zoom 1 to 23 times) is taken as the core data acquisition equipment (DAE) to construct a complete monitoring system featuring automatic target recognition, coordinate transformation models, image processing workflows and “end-edge-cloud” collaborative architecture, which realizes the automatic extraction of target features from images with a resolution of 8 million pixels, and intelligently obtains the vertical displacement of the target (pavement). The monitoring accuracy and reliability of this technology are verified through indoor and outdoor experiments, which can achieve the automatic and rapid acquisition and early warning of multi-point settlement data on the pavements under the advantage of low-impact traffic.
