Modeling Unordered Point Data for 3D GIS Mapping 2 approaches for improved usability and accuracy
The Problem GIS rendering increasingly powerful/flexible Precise, automated laser (LIDAR) point cloud data collection increasingly common This benefits engineers and archaeologists interested in highly detailed documentation Common scenes still challenging in GIS constraints in surface modeling, complexity Important for archaeologists to incorporate buildings, caves, viewsheds into GIS environment.
Current Approaches True 3D in ESRI applications Point cleaning and thinning
The Alternatives This paper presents two alternatives to 3D modeling in the GIS, which facilitate the use of these types of models. Modeling with 3rd party reverse modeling software Converting and cleaning laser scanner data files
Project Goals Make laser point cloud data usable and practical in the ESRI GIS environment Priorities Make files formats readable by GIS True 3D surfacing (as opposed to 2.5D) Georeference model Integrate with other GIS layers in ESRI GIS For point data, remove outliers For point data, thin points by 50% for import
Unordered Point Cloud Data point entities (x, y, z values) lack topology- no network between points can be imported into GIS and CAD as text files, and plotted as an event theme based on their provided coordinates
Unordered Point Cloud Data Very common laser scanning file type X, Y, Z, brightness No topology
Two Approaches Surfacing Complex Scenes Point Cloud Cleaning RapidForm, a reverse modeling software application Several datasets - very simple (a statue), to complex (a large viewshed) processing of the simple dataset most successful Point Cloud Cleaning custom application using Visual Basic and SQL focused on the basic goal of file selection, conversion, and decimation based on user input
Two Approaches Point cloud Each approach involves independent processing AND some level of import and integration in ArcGIS clean import import clean thin Cleaning Tool decimate triangulate Export cloud 3D Modeling Software Fix normals GIS import Fill holes import Interpolate Restore detail Export model
Surfacing Complex Scenes Open point cloud in modeling app The “overhang” problem What is “real” 3D?
Surfacing Complex Scenes Remove excess/ unwanted points Point cloud assessment and decision-making Data reduction
Surfacing Complex Scenes Clean Triangulation Normals Surface repair
Surfacing Complex Scenes Surface Surfacing Complex Scenes Thinning Triangulation Holes…
Surfacing Complex Scenes Model to cloud fitting Point density Subdivision Curvature/ heterogeneity
Surfacing Complex Scenes Sharpening Point Cloud Fitting Smoothing Spikes
Surfacing Complex Scenes Enhancement and analysis, etc. Enhancement Curvature Mapping Face Angle Plot Contours
Surfacing Complex Scenes Convert Surfacing Complex Scenes File Conversion Large .dxf files Normals display in 3D viewers Scale defaults
Surfacing Complex Scenes Open .dxf in GIS (ArcMap) Perform scale/geographic transformation Georeferencing Coordinate shifts Display bugs Rendering
Surfacing Complex Scenes Save as shapefile Open .shp in GIS (ArcScene) Converting to native format Dxf Shapefile Coordinate transformations Rendering
Surfacing Complex Scenes
3D Modeling Workflow Clean Surface Convert Open point cloud in modeling app Remove excess/ unwanted points Surface (Perform geographical Transformation) Fit surface to point cloud Convert (Perform geographical Transformation) Decimate by 75% Enhancement and analysis, etc. Open .dxf in GIS (ArcMap) Open .shp in GIS (ArcScene) Perform scale/geographic transformation Save as shapefile
Point Cloud Cleaning .pts file format and structure 459875 520.372084, 515.288156, 99.388029, 0.436561 520.373291, 515.290006, 99.387418, 0.434607 520.374988, 515.292604, 99.386801, 0.435218 520.377730, 515.296775, 99.387766, 0.441382 520.380551, 515.301068, 99.388923, 0.437049 520.380865, 515.301611, 99.386280, 0.432776 520.383698, 515.305925, 99.387768, 0.433265 520.385852, 515.309232, 99.387553, 0.434424 520.384259, 515.306885, 99.383045, 0.426490 520.385581, 515.308918, 99.382256, 0.418433 520.388582, 515.313481, 99.383157, 0.428321 520.391963, 515.318603, 99.385349, 0.430091 520.394570, 515.322571, 99.385707, 0.430335 520.397447, 515.326945, 99.387088, 0.434058 520.400273, 515.331245, 99.388221, 0.439063 520.402050, 515.333981, 99.387538, 0.454139 520.401986, 515.333934, 99.385228, 0.444007 520.405433, 515.339179, 99.386900, 0.443519 520.405661, 515.339573, 99.384751, 0.434180 520.407457, 515.342322, 99.384453, 0.428138 520.407630, 515.342620, 99.381949, 0.426795 520.410065, 515.346326, 99.382690, 0.429786 520.411375, 515.348333, 99.381632, 0.427771 520.414465, 515.353027, 99.383282, 0.429175 520.417455, 515.357582, 99.384478, 0.428565 520.420447, 515.362137, 99.385814, 0.429358 520.422668, 515.365539, 99.386229, 0.428321 520.423266, 515.366498, 99.384332, 0.428870 520.427095, 515.372303, 99.387053, 0.417212 520.428467, 515.374430, 99.385991, 0.419898 520.429136, 515.375478, 99.384438, 0.429236 520.430916, 515.378189, 99.384098, 0.429480 520.433170, 515.381632, 99.384570, 0.424598 520.434620, 515.383847, 99.384088, 0.418006 520.437265, 515.387864, 99.384658, 0.430030 .pts file format and structure Access database, template table Error-checking and data import
Point Cloud Cleaning Point cloud spatial distribution Density Brightness value
Point Cloud Cleaning CloudBurst point cloud cleaner File format Statistics Cleaning Decimation
Point Cloud Cleaning CloudBurst point cloud cleaner File format Statistics Cleaning Decimation
Point Cloud Cleaning CloudBurst point cloud cleaner File format Statistics Cleaning Decimation
Point Cloud Cleaning CloudBurst point cloud cleaner File format Statistics Cleaning Decimation
Conclusions Laser scanning devices/LIDAR aid archaeologists and engineers in mapping complex features So-called 3D GIS is not truly 3D enough to model laser point clouds effectively Reverse modeling software produces superb, intricate models and has overlapping file formats with GIS Expensive, difficult Numerous gotchas related to scale and coordinates Rudimentary point cloud cleaning and conversion software facilitates point cloud import and modeling in the GIS Simple models only! Decimation routine unacceptable- is there a solution?