Abstract: The traditional deformation monitoring method for primary support settlement in tunnels has many disadvantages, such as its dependence on single point measurement, sparse deformation monitoring points, lack of visual effect, and difficulty in finding deformed areas without monitoring points. Therefore, in this study, a computer vision image pointbased multidimensional deformation monitoring method for monitoring the tunnel primary support is proposed. It can be used to monitor and analyze the threedimensional overall deformation of tunnel primary support and twodimensional deformation of the crown settlement, perimeter convergence, and typical fullface deformation. The applicable results of the method show that the proposed method can comprehensively and intuitively obtain the deformation characteristics of the tunnel, and it is a supplement and improvement to traditional monitoring methods and provides a solution for tunnel construction monitoring. The deformation monitoring results can provide detailed data for studying the deformation law and characteristics of tunnel surrounding rock.

Key words: tunnel primary support, deformation monitoring, computer vision, image point cloud, crown settlement, perimeter convergence