There are a large number of redundant data in the point cloud acquired by 3D laser scanning method，thus，the point cloud data are needed to be simplified，while the original characteristics of point cloud data are utmostly reserved.A simplified algorithm based on local point cloud and global point cloud is proposed.The algorithm adopts non-uniform grid method based on point cloud segmentation to extract the local point cloud characteristics and adopts space voxel method to sample global point cloud.Then the two features are fused to obtain the optimal simplified characteristics.Experiment results show that this algorithm can meet the simplification requirements for various types of surface data.The maximum error of the point cloud simplification is 0.02812，and the average error of point cloud simplification is 0.000472.Comparing with non-uniform grid algorithm and space voxel method，this algorithm has high simplification efficiency and reserves the detail characteristics of the original data.