Contest No. 1


3D Modeling of Building from GeoSLAM Point Cloud Data


Description:

3D registration and modeling of 3.5 million raw point cloud data of Central Building of College of Engineering, University of Tehran (Figure 1, https://goo.gl/maps/1e68TswTfiYBni1h6). The point clouds are captured by handheld laser scanner GeoSLAM (two laz files). The final objective of this contest is to produce 3D model of the building including roof and wall elements (LOD3). Accordingly, the individual steps of the point cloud processing could be as follows:

a-      Preprocessing of point cloud: including cropping, noise reduction, filtering, point reduction and so on.

b-      Registration of both point cloud datasets.

c-      Geo-referencing of registered point cloud by GCPs.

d-      Segmentation and extraction of individual elements: planar or linear elements (depending on the employed algorithm/workflow) including roof faces, walls, dormers, ridge and eave lines, and so on.

e-      3D LOD3 model generation in vector format (shape files).

f-       Collecting some RGB images using ordinary camera from building and adding as texture to the final 3D model (optional).

g-      Produce final result as KMZ format which can be later exported in Google Earth platform.

h-      Writing a technical report by describing all operational individual steps

                            

Figure1: Central Building of College of Engineering, University of Tehran (©Google Maps)


                                                                                          

Figure 2: The raw point clouds collected by GeoSLAM (Courtesy: N.P.R. Ltd)


Deliverables:

After producing the final LOD3 model, the following materials should be provided and send it to the conference email address in zip format:

-          Technical report in docx or pdf describing all detail analysis (NOT more than 5000 words)

-          Produced model in shape format (.shp): the shape file should be opened in any standard GIS/Photogrammetry software. It can be also including the textured images of the building faces.

-          KMZ file

-          Programming codes (if employed)


Evaluation Criteria:

The following criteria are considered for the evaluation of the accomplished tasks:

-          LOD3 with the maximum geometric details.

-          Geometric accuracy of the extracted details.

-          Realistic representation of 3D model by texture mapping and lighting.

-          To develop and use your own program code (preferably open source) instead of using existing commercial tools.

-          A comprehensive technical report (in Persian or English).

-          Level of scientific innovation.


Data availability:

The web link for downloading the point cloud data will be available upon the completion of the registration form.