Laser scanners
Selecting and validating laser scanning systems for dimensional inspection, with clear limits on density, surface response, registration and uncertainty.
Tags: inspection, dimensional-metrology, laser-scanning
On this page
Verify the scanned coverage and alignment before using a point cloud or mesh for dimensional inspection.
Illustrative sampling example: ideal parallel profiles spaced 0.20 mm apart can entirely miss a 0.10 mm-wide groove running between them. More points along each profile do not close that cross-profile gap. This is a geometric example, not a scanner resolution claim or universal minimum-feature rule. Actual detection also depends on beam footprint, incidence, surface response, filtering and viewing direction.
Measurement families
| Family | Principle and selection implication |
|---|---|
| Laser triangulation | A camera detects a projected laser spot or line; calibrated geometry determines surface position. Moving the sensor or part extends coverage. Common for component inspection. |
| Ranging/terrestrial scanning | Range and angular measurements locate points over a larger scene. Time-of-flight and phase-shift implementations have different error characteristics from close-range triangulation. |
| Structured-light scanning | Projects patterns over an area and observes their deformation. It may also use triangulation, but is not interchangeable with a laser-line scanner merely because both produce point clouds. |
ZEISS describes the laser-spot/line and projected-pattern approaches and emphasizes a stable installation when motion extends laser coverage. Choose the actual optical principle, measurement volume and feature capability; “3D scanner” is too broad a specification. ZEISS, Understanding Triangulation, undated technical page
NIST's 2021 review covers terrestrial scanners, including time-of-flight/phase-shift ranging, angular errors, environmental influences and registration. Its large-volume findings should not be transferred numerically to a handheld triangulation head. Muralikrishnan, Performance Evaluation of Terrestrial Laser Scanners—A Review, 2021
Scanner performance versus system performance
Identify how the scanner pose is measured: an articulated arm, CMM, tracker or another tracking/registration arrangement contributes to the result. Specify the complete configuration and working volume. A sensor-head test does not establish every length measurement after movement and registration.
A bounded example comes from Hexagon's Absolute Arm brochure, copyright 2024. It lists the AS1 head's minimum point spacing as 0.027 mm and its accuracy entry as 0.013 mm, using the stated PForm.Sph.1×25:ODS test designation. Separately, its 8520-7 arm + AS1 system has SSA 0.039 mm; the 8545-7 + AS1 entry is 0.134 mm. The brochure defines SSA through LDIA according to ISO 10360-8 Annex D. These are different tests/configurations, not numbers to substitute for one another or an uncertainty guarantee for every feature. Hexagon, Absolute Arm brochure, specification spread 32–33
Use the supplied specification and calibration/acceptance report to identify test artefacts, metrics, conditions and exclusions. Keep the scanner, carrier, software and relevant settings identifiable when comparing systems.
Point spacing, coverage and hidden geometry
Point spacing describes sampling density, not the maximum error of a coordinate or fitted feature. More points can describe a surface more densely while retaining systematic errors. Acquisition rate likewise describes throughput rather than dimensional correctness.
Plan views for the feature, including its edges and datum surfaces. Laser illumination and camera reception must both reach the measured region in a triangulation setup; a visible entrance does not establish coverage of a deep recess. Inspect missing data before meshing. Interpolating a hole in the mesh creates an estimated surface, not a measurement of previously hidden material.
Materials, orientation and environment
Qualify the actual finish, colour and geometry. ZEISS notes that highly shiny or translucent objects can remain difficult for laser triangulation. A scanner's advertised ability to capture reflective parts does not establish performance for every polished edge or transparent surface. ZEISS, laser-triangulation limitations
NIST's 2011 experiments on a laser-spot triangulation probe investigate otherwise hidden performance effects, including material response, probe orientation and near-grazing measurements. Their results demonstrate why a nominal sensor specification is insufficient for an arbitrary surface; the measured errors belong to the tested probes and conditions. Validate representative angles and surfaces rather than borrowing that paper's numerical errors as a correction for another scanner. Muralikrishnan and colleagues, Performance Evaluation Experiments on a Laser Spot Triangulation Probe, 2011
If matting spray is needed, record and qualify it. FARO warns that dimensional degradation depends on how much and how evenly material is applied. Its discussion cites a particular powder study; do not treat its estimated coating thickness as a universal compensation. The measured surface may be the coating rather than the original part. FARO, Scanning Accuracy effects on Parts When Using Powder Spray
Control mounting stability, relative part motion and thermal conditions. Large-volume work additionally needs attention to the propagation path and registration, as described in NIST's terrestrial-scanner review. Operating-temperature permission alone does not demonstrate measurement uncertainty throughout that range. NIST, terrestrial-scanner review
Registration, alignment and feature evaluation
Registration combines scan views. Inspection alignment places the measured data in the reference system used for evaluation. Keep these purposes explicit: a small overlap residual does not independently prove dimensional truth or a correct datum reference frame.
SHINING3D Inspect 1.2.0 documentation distinguishes global best fit, local alignment and datum alignment. Best fit minimizes overall average distance deviation; local alignment uses selected faces; datum alignment uses the chosen datum features to constrain the model. Thus changing alignment can change a deviation map. Select the method from the inspection requirement and retain its settings, rather than choosing the most favourable colours. SHINING3D Inspect, Alignments, version 1.2.0
Document feature extraction, excluded points, filtering, mesh reduction and fitting method. For a bore or edge, verify that sampled geometry supports the required measurand; a fitted cylinder from a limited visible patch needs separate validation.
Validation and reporting
Evaluate the task's uncertainty, including reference artefacts, system errors, surface interaction, registration, repeat positioning and data evaluation. Repeated scans can reveal variation while leaving a common bias unchanged. NPL distinguishes resolution, repeatability, calibration and uncertainty and explains reference-based verification; use appropriate artefacts and representative measurements, not point count, to establish confidence. NPL, Fundamental Good Practice in Dimensional Metrology, GPG80, July 2005; corrected reprint October 2012
Retain the system configuration, raw-data coverage, preparation, scan settings, alignment, feature definitions and applicable checks with the results. Report uncertainty and the agreed decision rule when judging conformance near a limit.