Point cloud data has the characteristics of measurability,high accuracy,fine details and fast acquisition.In order to build the ancient building model with point cloud data,with the color 3D laser point cloud data of ancient buildings as the research object,the method of rapid automatic classification and target extraction of ancient building point cloud is proposed.Firstly,the original point cloud is transformed into raster data,and the point features and global features at different scales are extracted by multi layer perception.Then,the MLP parameters are optimized by particle swarm optimization algorithm,and the automatic classification and extraction of point cloud data are realized.Finally,the point cloud data of an ancient building collected by Z + F5010C scanner is taken as the test object to verify the feasibility and practicability of the algorithm,laying a foundation for realizing the ancient building′s parametric modeling in Revit and realizing the unified management based on the information platform.