Laser scanners allow surveyors to create precise 3D models with relative ease. To make sense of that data, point clouds are used to represent the information and then align scans for the creation of composite surveys. The process of identifying and aligning scans is called ‘registration’. Traditionally, this has been one of the slowest and most time-consuming aspects of creating a point cloud survey, requiring lengthy manual procedures in both the field and the office to place all of the collected data in a common context. This article will explore these traditional processes along with advances that are now offering a better way — enabling surveyors to process and register point clouds faster.
The catch 22 of point cloud registration is that efforts to save time in the field often result in time costs in the office. This issue typically revolves around target placement and the debate over targetless registration.
The traditional method of registering scans is to use artificial targets. Two-three point reducible objects (either free-standing spheres or chequerboard surface targets) are placed within areas of scan overlap and used to ‘fix’ scans in a common context. This is a robust solution, but it is manual and time-consuming.
Care must be taken to ensure the appropriate placement of each target. For surface targets, subsequent scan angles must also be accommodated. In both enclosed and open areas, the target range needs to be considered based on the type of targets, their size, the relative location of the scanner and scan resolution settings. This is not particularly hard, but it takes time.
For years, surveyors have had the option to avoid this by using natural features within the scene to align scans. The problem with this is that it takes far longer for processing software to identify and overlap scans using this method. This wouldn’t be so much of a problem other than the fact that a lot of processing software requires much of the operation to be manually overseen. This creates a significantly more time-consuming procedure that reduces any efficiencies gained through avoiding target placement while scanning.
Speeding up the registration process essentially comes down to software. Improving this process is how surveyors can create efficiencies. If you can find software that mitigates some of the time costs of approaching registration without targets, you can avoid target placement and minimise the consequences.
Software that solves this issue has appeared in the market. The solution is multi-stage vector analysis. This cuts processing into three distinct stages (rotational, horizontal and vertical alignments) to deliver a faster but more robust procedure.
To achieve this, the positional data of the scanner is used to extrapolate each point into a directional vector with a normalised unit length of one. This allows each scan to be condensed into a single point, creating a ‘vector sphere’. The density and directional characteristics of the vectors retain the unique identity of each scan. Representing scans in this way allows for easy rotational alignments through placing adjacent ‘vector spheres’ within one another. Once rotational alignment is achieved, vertical and horizontal alignments can be approached using rapid 2D point density assessments.
Depending on the size of the dataset, this can deliver an efficiency improvement of 40%-80%. Critically, the scrutiny with which each distinct stage can be approached removes the need for cross-checks during processing and limits the requirement to set scan parameters. This automates some tasks and allows the remaining manual operations to be front-loaded or relegated to post registration review.
This combination of techniques not only reduces the total time costs of processing targetless data, it completely mitigates what inefficiencies remain by allowing scans to be queued up for hands-off processing. In fact, multi-stage vector analysis will deliver in-office processing efficiencies when compared to the capabilities of traditional software to handle data fixed by artificial targets.
Point cloud processing and registration are the stumbling blocks to efficient point cloud creation. This has always come down to software. The ability to align scans using natural features has been around for some time, it has simply never been efficient. By approaching scan alignment in three distinct stages, rather than attempting to look at the data holistically, the most advanced processing software is capable of delivering aligned datasets faster and with less manual involvement. Surveyors who embrace this technology can improve their workflows and significantly increase the efficiency and speed of their point cloud processing and registration.