This paper examines the procedures on reduction point data acquired by laser scanning device. From the evaluated procedures, applying point sampling and filter redundancy method are very useful for both freeform and primitive shape point cloud. From the tested procedures, applying filter redundancy on sampling point data can shrink the point cloud amount with good accuracy while maintaining the quality of the measured surface. The surface model generated from this proposed reduction procedures can be used in further modeling processes (e.g. registration, meshing and segmentation) in order to generated fine 3D model and perfect CAD model. For further research, the point reduction will be tested on 3D model generated from merging the mesh polygon using the procedure in this study