Abstract: [Purposes] Aiming at the problems of complex construction site environment and low efficiency of manual inspection in the pre-installation process of steel structure buildings,this paper proposes a hierarchical ICP point cloud registration algorithm based on the engineering background of Taiyuan Airport Terminal.[Methods] This method stratifies the point cloud along the main direction and performs ICP registration layer by layer,and adaptively selects the optimal matching result,which solves the problem of low point cloud overlap rate and point cloud registration failure caused by environmental constraints in the data acquisition process.In order to quickly detect the error value of component bolt hole products,based on the point cloud main plane projection and two-dimensional morphological analysis of hole characteristics,an automatic bolt hole recognition method based on image processing mechanism is proposed.[Findings]The results show that the matching degree of the hierarchical ICP algorithm is more than 15% higher than that of the traditional algorithm,and the accuracy of bolt hole recognition is more than 94%.[Conclusions]This method has been verified in the pre-installation stage of steel structure,which can provide data support for the site and significantly improve the construction efficiency and assembly accuracy．

Key words: 3D laser scanning technology, steel structure detection, point cloud registration, target recognition