Direct Answer
How to Review Laser Scan Registration Quality
Registration quality cannot be reduced to one green badge or average error. Review the scan-link network, the geometry or targets supporting each connection, survey-control agreement, loop closures, inclinometer use, local cross-sections, representative dimensions and coverage. Acceptance criteria should be chosen for the project's intended use and documented with the delivered data.
Start with the registration network
A scan network shows how stations are connected. Look for isolated stations, long chains, single bridge links and zones without loop closure. A redundant network offers independent paths that can reveal disagreement; a fragile network can produce a plausible overall shape while allowing error to accumulate through the project.
The links should correspond to actual site relationships. A connection across non-overlapping rooms, repetitive structural bays or moving geometry deserves investigation even when software accepted it automatically.
Inspect the evidence behind each relationship
For target-based work, verify target identity, distribution, observation angles and residuals. Targets concentrated in one plane or one small part of the view may constrain position less effectively than references distributed around the scanner. For cloud-to-cloud work, inspect overlap and the distinctiveness and stability of the common geometry.
FARO SCENE marks correspondence quality and provides tools to find, clear or force correspondences. A forced relationship should be treated as an operator assertion that requires stronger visual confirmation, not as a shortcut around missing overlap.
Understand point distance and tension in SCENE reports
FARO describes Point Distance as the standard deviation of distances between local reference points and their corresponding references. It describes agreement among the references used for a scan; it is not a direct statement that every cloud point is within that value of truth.
The report describes Tension as discrepancy between corresponding reference objects in the global coordinate system. Position contributes for point references, while position and direction can contribute for planes, slabs or pipes. Values closer to zero indicate better correspondence agreement, but the acceptable value still depends on reference type, geometry, units and project requirements.
| Diagnostic | What it helps reveal | What it cannot prove alone |
|---|---|---|
| Network graph | Connectivity, redundancy and isolated zones | Local surface agreement or site coverage |
| Point distance / residual | Agreement among used reference observations | Accuracy of every point or modeled element |
| Tension | Discrepancy between corresponding reference objects | Correct target identity or adequate overlap |
| Control residuals | Agreement with known coordinates at control observations | Quality between sparse control points |
| Visual section | Local double surfaces, offsets and edge mismatch | Global coordinate agreement across the whole project |
Use survey control without forcing the network
Control can establish coordinates, orientation and independent checks. Confirm units, point IDs, axis order, coordinate reference and whether control is held fixed or weighted. Review both fit to control and the internal scan network; forcing a weak network onto control can distribute distortion between controlled points.
Whenever possible, reserve independent checkpoints that are not used to solve the registration. A check that participates fully in the adjustment is not an independent verification of the result.
Look for drift and local mismatch
Inspect loop closures, long corridors, vertical circulation, repeated bays and the transitions between separately processed clusters. Slice the cloud through flat floors, walls, columns, pipe runs and edges. Double surfaces, thickness changes or rotating offsets can reveal a problem hidden by an overall 3D view.
Use stable features and understand their surface response. Glass, water, vegetation, moving vehicles and highly reflective or absorptive materials can create noise that should not be treated as a reliable registration reference.
Separate registration quality from coverage and accuracy
A well-registered project can still miss the back of equipment or the top of a beam. Coverage review asks whether the required surfaces were actually captured. Accuracy review asks whether the complete workflow meets the project's stated acceptance method. Registration diagnostics support that judgment but do not replace it.
Connect checks to the intended use. A background coordination cloud, a fabrication tie-in and a survey-controlled deformation study require different evidence and may require different instruments, control and procedures.
Document the approval
Record software version, registration methods, control basis, excluded scans, major corrections and the person responsible for approval. Export the relevant SCENE report or QA summary and pair it with a file manifest and coordinate notes.
A recipient should be able to understand what was checked and repeat key validations after import. That is more valuable than a screenshot of a single summary value without context.
Approve a registration through multiple independent views of quality: network, references, control, local geometry, coverage and downstream coordinate checks.
Frequently asked questions
Primary sources and standards
Technical claims in this guide were checked against the following manufacturer documentation, standards bodies and software documentation.
- SCENE registration report informationFARO Technologies · opens in a new tab
- Manual registration correspondence toolsFARO Technologies · opens in a new tab
- Registering with survey data in SCENEFARO Technologies · opens in a new tab
- FARO SCENE software technical sheetFARO Technologies · opens in a new tab