3D laser scanning is a rapid, non-contact, accurate and objective method for digital documentation of the built environment; urban streetscapes, buildings, heritage structures, archaeological sites and terrain.
Laser scanning allows us to accurately record the 3D surface geometry of objects in a digital form. A large amount of 3D data can be collected within a short period of time. Scanning can be done from a static position on the ground or from aircraft – when it is normally called LiDAR (Light Detection And Ranging).
The laser beam scans the object surface up to 1 million times every second. The returning, reflected laser light is used to compute the distance to the surface. This means up to 1 million spatially accurate coordinates (or points) are collected every second. The millions of points computed are used to produce a ‘point cloud’ which defines the surface geometry of the building or monument of interest.
The data generated by the laser scanner is a ‘point cloud’: a collection of xyz coordinates, which may include other information (e.g. intensity of return of laser signal or RGB value). Point clouds are acquired from several points of view and joined together to produce a 3D image with accurate dimensions. The point cloud can be useful in itself, or it can be further processed to produce models, photo-realistic surfaces or outputs in other forms (e.g. for AutoCAD, 3D printing or milling machines).
3D objects need to be scanned from several angles to obtain the optimum amount of data. The number of scans required at each site depends on the intricacy of the object; complex objects require more angles of view to ensure that important detail is not missed. There needs to be sufficient overlap between scans to enable them to be joined together into the 3D model.
In most cases surveying targets are used to join individual scans. Ideally, at least four should be visible from each scan position to enable the scans to be ‘registered’ together with a good degree of accuracy. In practice point clouds can also be registered by using points, edges and surfaces in common.
The final registered point cloud will be based on an arbitrary reference point (0, 0, 0 with reference to the first scan position). If the geographical position of the object needs to be known (i.e. it must be ‘georeferenced’), then points within the scan need to be surveyed (e.g. using a GPS or GNSS system). The scan can then be positioned accurately within a known grid system.
Laser scanning accuracy and resolution generally depends on the scale of the surface being recorded. Small scale objects (of the order of a metre of so across) can be recorded with a very high degree of accuracy (to 10s of microns); building facades and the like with an accuracy of 4-6mm. This variation in accuracy is because different scanning technologies are required at different scales.