Abstract: In order to aquire the global and local deformation of primary support rapidly and accuratly, an image point cloud space distance measurement algorithm is put forward based on image sparse point cloud and dense point cloud of tunnel primary support obtained by computer vision algorithm and the characteristics of the global model based on Hausdorff distance and local mode based on least square fitting plane. By using the algorithm mentioned-above, the smaller distance calculated by the global model and local model can be reserved for each point, which helps to solve the problems that global model has high requirements on point cloud density and local model has large deviation in local fitting plane; further more, the direct comparison among multi-stage tunnel image point clouds can be realized, and the calculation steps and post-processing process can be simplified, which improves the monitoring speed and accuracy. The algorithm is applied to the global deformation monitoring of primary support of section ZK61+990~ +994 of left line of Baiyanzi Tunnel on Shangri-la-Lijiang Expressway in Yunnan, and the monitoring results show that the algorithm can directly and visually reflect the overall deformation of the tunnel, and the calculation results are accurate and reliable.

Key words: tunnel construction, primary support deformation, monitoring technology, image point cloud, distance
algorithm, mathematical statistics