Registration of point cloud data is an essential part of the processing workflow and determines the resultant quality of the combined data. However, it is only as good as the data captured.
In the traditional surveying workflow the scanning instrument is placed on a tripod on several pre-determined stations to ensure good coverage. The process can be combined with total station measurements to obtain control information for the tie points, measure the position of the stations or a combination of both.
When scanning, especially with an automated registration workflow in mind, there are some simple rules that are useful to follow to ensure that automated registration runs as smoothly as possible:
- To be most effective with automatic registration software, such as Vercator, requires thought about the capture process. The user capturing the data should think about the geometry of the scene as natural targets in the same way as with mobile mapping SLAM-based systems. This means that higher geometric variation in the scene will aid the process greatly and ensure the most successful automated result.
- Ensuring reasonable overlap between scans of around 60%. There is some tolerance to this in the case of the amount of unique geometry in the scene where it can be lower in some circumstances.
- Minimising acute angles between scans. Laser scanners generally provide poor quality returns from surfaces that are at an acute angle to the laser beam. Therefore, it is important to minimise these in the scene where possible especially in overlap areas between scanner positions.
Tags: pointcloud
