Autodesk Research
Blog

The 3D scanning process
April 29, 2009

The first digital dataset we want to provide for 210 King Street East is a 3D laser scan of the office. As the name suggests, laser scanning employs a laser to determine the distance from the scanning device to surfaces around it. This real-world spatial data is collected as points in 3D space relative to the position of the scanner. Carrying out a full scan of the entire building would have been a huge task, so we limited the scope of the scanning to the entire fifth floor and the rooftop terrace, as well as the lobby and the exterior of the building. In total, the dataset contains over 1.3 billion data points.

In support of this project, Faro Technologies graciously provided us with a Faro Photon 80 laser scanner. This self-leveling, tripod-mounted unit is capable of scanning a radius of 0.6 to 76 meters, with a precision of 2mm. Since a laser scanner can only capture data in its line of sight, it is necessary to take multiple scans of a given room to eliminate the effect of obstructing features, such as pillars, doors, and desks.

To facilitate merging the data from these individual scans, we need to identify unique features in each of the scans that relate to each other. We accomplished this through the use of position registration markers. We spent the first morning setting up these markers around the office. When positioning these, we had to ensure that any given scan would have line of sight to at least 3 of these registration markers. A total of 61 registration markers were positioned.

The next step was to scan the office. Using a floor plan, we figured out the best place to position the scanner to minimize the number of scans required, which also reduced the amount of redundant data captured. Each scan took approximately 5 minutes. Over the next day, we carried out 53 individual scans of the office.

Simultaneous to our scanning efforts, surveyors from Bennett Young were determining the terrestrial space position of the registration markers in terms of coordinates and elevation above sea level. This will allow us to know the precise location of any 3D point in our dataset in terms of the real-world. The surveyors measured the registration points on both the interior and exterior of the office, correlating the interior measurements to exterior street measurements from the deck.

We have just started post processing of the data. Here is a sample of the point data from one scan – approximately 27 million points – the spherical data is shown as a flattened image. Registering each scan to the other is an interactive process whereby the ID numbers of the registration markers need to be manually confirmed. Stay tuned; more data will be posted shortly.

ZIP
The 3D Scanning Process High-Res Images
ZIP - 22.173MB


Previous | Next