Abstract: In the process of digital mine construction，3D laser scanning is used to obtain point cloud data of surface or buildings quickly.The point cloud boundary is not only an important geometric feature to represent surface，but also serves as a model to solve the domain of the surface，which plays an important role in reconstructing the quality and precision of the surface model.It is necessary to extract the boundary points from the mass data in order to model the laser point cloud data.In this paper，a method is proposed to automatically extract the boundary features of the point cloud by discussing the geometric distribution of the local point set projection onto the micro-tangent plane fitted by them.This algorithm could be applied to various point cloud data in complicated surface with its rapid processing speed and accurate extraction result.

Key words: Digital mine, 3D laser scanning, Boundary point, Feature extraction, K-nearest neighbor