Digitally Capturing the Quon Sang Lung Laundry
A multi-sensor approach involving both UAV photogrammetry and terrestrial laser scanning (TLS) was used to capture the interior and exterior of the laundry building. Drone-based photogrammetry was used to capture the building in its current urban context, capturing its location relative to other nearby buildings. A DJI Phantom 4 drone was flown in a double grid pattern over the laundry shop at an elevation of 40 meters at a camera angle of 70 degrees, resulting in an overlap of 90%. The resulting images were processed using Pix4D mapper software and used to create a point cloud, mesh and final 3D model (Figure 5 and 6). Photogrammetry is an excellent way of creating visual models of artifacts and buildings. However, the accuracy and precision of these models is often too coarse to allow for the calculation of volumetric and linear measurements, which is necessary for the construction of the 2D architectural plans. Terrestrial laser scanners (TLS) can provide point clouds with millimeter and submillimeter accuracy.
Two laser scanners were used to record the building – a Z+F 5010X scanner, and a Faro Focus 3D scanner. A series of paddle, spherical, and paper targets were placed at various points around the structure, and well as within the main and top floors of the building (Figure 7 and 8). To ensure accurate point cloud registration, a minimum of three targets need to be common to two or more overlapping scans. Targets are not always necessary with the auto-registration algorithms found in programs such as Leica Cyclone, but they increase accuracy. Since one of the goals of the project was to create a set of architectural plans, the use of targets for registering the data was an essential part of the workflow.
A total of 16 high resolution scans of the exterior and interior of the building were completed over a period of 3 hours, during light snow and a temperature that hovered around 0 Celsius. Detailed recording of the structure using hand measuring tapes, as is the common practice for recording historic structures, would have taken much longer. Furthermore, the resulting data would have likely been prone to human errors in measurements.
One of the advantages of the Z+F 5010X scanner is that it allows the operator to register the point cloud data in real time, as the scanning is performed. This is advantageous because it allows the operator to identify if there are any gaps in the data – in other words, the operator knows immediately if a critical area or feature of the building has been missed by the scanner. It also speeds post-processing of the data upon returning to the office. Fine registration of the scans was completed using Leica Cyclone and Z+F Laser Control software.