Aiming at the existence of various morphological noise points and a large amount of redundant data in the original point cloud scanned in the field, this paper proposes a simplification optimization strategy for point clouds based on comprehensive algorithms such as method library, cloth simulation filtering, and curvature classification. First, sparse noise points at long distances are removed by statistical filter. Second, passthrough filter is used to segment point cloud blocks with close distances and large density , and cloth simulation filtering algorithm is used to remove such noise points, and then using radius filter to remove the close distance noise points around the target point cloud. Finally, the redundant data of the point cloud is removed based on curvature-grading compression method and compared with two traditional compression methods for experimental comparison and analysis. Experimental results show that the simplification optimization strategy proposed in this paper can effectively remove the noise points in the point cloud, while retaining most of the characteristic points of the point cloud, it can minimize the redundancy of the point cloud data and improve the data quality of point cloud model reconstruction.