In this article we want to talk about a laser scanner survey carried out with the technique of target acquisition directly in the field, a working methodology that is certainly more “onerous” in terms of time in the “on site” survey but extremely faster and more accurate in the subsequent work of cloud merging.
We used this methodology of work during the general survey of the Pieve di Corsignano Church carried out by La SIA to support the crack framework survey.
Several stations were carried out both outside and inside the facility, specifically 13 outside stations and 12 inside stations for a total of 25 stations completed in about 7 hours of work. The instrumentation used was a Leica P30 Laser Scanner and 6 4.5-inch black-and-white type targets (4 mounted on swivel mounts and 2 fixed on the wall); Leica Cyclone software was used for cloud merging.
The layout of the church is rectangular with a bell tower in the lower left corner and an attached sacristy in the upper right corner; on the outside of the church, behind the altar, are the remains of what must have been the bases of the church apse.
The survey first involved the exterior: starting from the northeast façade, the 4 sides were turned counterclockwise to the part towards the southeast where the outcropping remains of the original wall layout (the subject of recent archaeological excavations that brought them to light) were also surveyed. We then moved to the interior by placing stations in the three naves, bell tower and sacristy (no underground crypt was surveyed), trying not to leave blind spots and giving more detail to some areas indicated by the structural engineer.
In surveys, especially of exteriors, we almost always use targets in order to have a better result in the subsequent phase of cloud merging; this time, given also the greater precision required, we also opted for the acquisition of targets directly at the end of the scan through a dedicated function of the scanner.
Each scan was linked to the next with at least 3 targets placed at different distances and different heights, always trying to have the fixed points not clustered close together but “scattered” in the scene to be acquired. The acquisition of the targets, with the laser scanner used (Leica P30 model), takes place after the scene has been scanned: without obviously moving the instrument, the visible targets are selected from the scan preview one by one, assigning each one a unique ID (in the survey notebook, where we usually print the floor plan of the building to be surveyed, along with the approximate position and the scan number we also mark the ID of the various targets placed).
During the survey, the correct assignment of the right ID to each target is critical so that you do not end up at the studio with clouds that do not lock (of course, the assigned ID can always be corrected later in the subsequent cloud processing phase). Once the targets have been identified and registered, we move on to actual acquisition by the instrument, which reactivates and searches the black-white pattern of the circular target at the previously selected points.
If the acquisition is successful (the instrument succeeds in locating and scanning the target), a positive check is obtained from the instrument on the detected targets and they can proceed to register them in the scene.
But what is the difference between acquiring targets in the field and doing the same operation in the studio on the single point cloud? Operationally speaking none, in either case you get fixed points in the scene to be used for merging the various clouds. Technically, however, there is a great deal of difference, mainly in the accuracy of target center relief, and we can tell this already from how the instrument “tares” when it has to acquire the target.
The Leica P30 has several preset resolutions for scanning (from a maximum of 0.8mm to a minimum of 50mm calculated at 10m), when it acquires targets “on site,” already knowing the distance at which the targets are positioned (a scan has already been made) it “tares” to get the right detail on each target; standing close to the instrument you can see (and feel…) the prism, which has a different rotation speed (the slower it is, the higher the scan resolution) depending on the distance of the target to be acquired.
This allows, for example, a complete scan of the scene to be made at a low resolution (e.g., 50mm) and, subsequently, an acquisition of a target as distant as 50m with a resolution of 0.8mm. On the other hand, if in cloud post-processing we were to try to acquire a target 50m away in a scene scanned at a good resolution (e.g. 3.2mm) we would have a possible error of the “detected” center of the target of 8mm (a bit too much for making a cloud merge where much precision is required).
This fact of the possible error is related to an aspect that should be taken into account, namely that the “on site” target survey, downstream of the scan, generates a new point in the scene based on the “second” scan of the target only (the “true” center is identified through an algorithm that takes into account the black and white portions of the target); on the other hand, in the target recognition procedure at the cloud post-processing stage, the algorithm always generates a new point in the scene (the center of the target) but based on the points already existing in it with all that follows in terms of accuracy.
-
Relief -
Zoom in on relief detail
We tried on a target only 11m away from the instrument to perform the two procedures i.e., onsite recognition and post-processing recognition; the scan resolution was 3.1mm@10m. As you can see in the image opposite, the error between the two detected centers (“01” is the center of the target detected on site and “C” is the center of the target detected in cloud post processing) was 1mm (0.98mm to be precise).
Finally, there is the aspect related to the actual points recorded in the scene and thus the weight in terms of GigaBytes of files to be managed. Doing high or very high resolution scans to record anything that “might” be useful (including targets) in the later post-processing stage makes little sense besides wasting a lot of time on site. Each scan or survey campaign must be “calibrated” according to the purpose of the survey itself and the subsequent return phase.
To recap if you want to achieve greater accuracy in joining stations in your surveys, we recommend that you try the technique of onsite target acquisition.
Good work
