Abstract: 3D laser scanning technology has the advantages of high efficiency,high precision and low cost,and can realize the rapid construction of complex mine 3D model. However,this technology has some problems such as large volume of point cloud data and large noise,which brings a lot of inconvenience to data storage,transmission and processing. To solve the above problems,a three-dimensional laser scanning point cloud composite filtering algorithm for roadway is proposed. In this paper,data filtering is divided into two processes:reduction and noise reduction according to the algorithm focus and point cloud characteristics,and then combined with the characteristics of mine semicircular arch roadway,a " profile-axial-random" composite data reduction filtering model and a composite data noise reduction filtering mathematical model are proposed,and the applicability,reliability,rationality and accuracy of the composite algorithm are compared and analyzed. The test results show that compared with the single algorithm,after using the composite filter algorithm to simplify and reduce noise,the original contour features and details of the roadway are reasonably preserved,the data simplification rate is as high as 99. 3%,and all kinds of outliers such as pepper and salt noise and Gaussian noise are filtered out to the maximum extent. It achieves the goal of large-scale data simplification and efficient processing of outlier points under the premise of ensuring model quality,and lays a foundation for real-time state display,dynamic evaluation and behavior prediction of mine whole life cycle.

Key words: 3D laser scanning,semicircular arch roadway,data filtering,point cloud data,mine intelligence