Laser triangulation scan spacing: prevent stretched clouds

A laser profile can be accurate across the light stripe while the assembled point cloud has the wrong length. The missing information is often the travel position assigned to each profile, not the triangulation calculation itself. This procedure is for engineers building or integrating a scanning station. You will choose a position source, calculate the required profile rate, associate motion with exposure, and verify the reconstructed scan axis against known dimensions.

1. Choose a measured position source

Treat every profile as a measurement with its own acquisition position. Assigning the next profile to the next equally spaced row is valid only when acquisition actually follows equal travel increments.

For continuous scanning with changing speed, position-based triggering is usually the most direct approach. An encoder measures travel, and compatible acquisition hardware requests a profile after each selected displacement. Confirm the complete trigger path in the equipment documentation; an encoder connector alone does not establish the required behavior.

For a prototype, calibrated stop-and-capture motion can be easier to validate. Move to a measured position, wait for motion to settle, acquire the profile, and record that position. Hand movement without position measurement cannot provide reliable scan-axis dimensions.

If the cloud slopes rather than stretches, use the separate laser-triangulation tilt calibration checks. Changing scan spacing does not correct a miscalibrated laser plane.

2. Set pitch, speed and exposure together

Define the scan axis as the direction in which consecutive profiles build the surface. Let the required pitch be the distance between profiles along that axis. This pitch is separate from the point spacing within an individual laser profile.

For constant speed, the basic relationships are:

Suppose the application moves at 100 mm/s and requires a 0.10 mm scan pitch:

The exposure movement is one tenth of the requested pitch. That is a useful starting comparison, not an accuracy guarantee. Surface slope, stripe extraction, vibration and calibration also affect the result.

The line-rate calculator covers the same speed-to-sample-spacing arithmetic. Use its numerical estimate for planning, not its camera suggestions as proof of profiler compatibility.

Choose pitch from the smallest relevant feature and the measurement tolerance. A 0.30 mm feature spans only three nominal intervals at 0.10 mm pitch, so its measured shape can depend strongly on its position between samples. Validate the chosen pitch using representative edges, grooves and surface slopes.

3. Connect acquisition to travel

Use the following sequence when the documented hardware supports encoder-based acquisition:

1. Establish the encoder scale in the receiving device's decoded counts per unit travel.
2. Confirm count direction by moving a known positive distance.
3. Select the travel increment that requests each profile.
4. Configure the documented trigger mode and acceptance limits.
5. Record profile identifiers, trigger information and position metadata where available.
6. Test at the highest intended speed before measuring parts.

An encoder's stated pulse count may differ from the decoded count used by the receiving device. Confirm whether its setting refers to cycles, edges or another documented convention.

For example, suppose the complete measurement chain produces 2,000 decoded counts/mm. A target pitch of 0.10 mm requires:

At 150 mm/s, the same pitch requests 1,500 profiles/s. The pitch remains unchanged, but the required acquisition rate rises. If a device cannot accept that rate, triggers may be rejected or profiles may be lost. Do not assume the resulting data still has consecutive, equally spaced samples.

Follow the hardware-triggered acquisition procedure when checking trigger configuration and frame handling. Position validity and frame delivery are related, but they are not the same test.

4. Match position to exposure, not arrival time

For free-running acquisition, sample position and image time in a shared or calibrated time reference. Use documented exposure timestamps and synchronized encoder data where available. A timestamp recorded when software receives a frame includes transport and buffering delay; it is not automatically the exposure time.

When position samples bracket an exposure timestamp, linear interpolation can estimate the acquisition position:

Here, the two position readings are s_a and s_b, recorded at times t_a and t_b. This estimate assumes approximately linear motion between those readings. Rapid acceleration requires closer position sampling or a better motion model.

A timing uncertainty also becomes a position uncertainty:

Use a consistent exposure reference, often the exposure midpoint when appropriate for the device and measurement model. A fixed trigger delay creates a position offset at constant speed; changing speed can make that offset vary.

For a camera-based prototype, shutter behavior matters too. A rolling shutter can expose different image rows at different times. Do not assign one instantaneous pose to the whole image without assessing the resulting error. Confirm shutter behavior and timing before selecting from area scan cameras.

5. Use calibrated motion when an encoder is unavailable

A commanded speed is not evidence that the object maintained that speed during every exposure. Acceleration, drive following error, slip and manual movement can all break the equal-spacing assumption.

Suppose software assigns 0.10 mm between profiles, while the object actually moves 0.08 mm between exposures:

The scan axis appears 25% too long. If actual travel rises to 0.12 mm per exposure, the ratio becomes 0.10 / 0.12 = 0.833, making that region approximately 17% too short. Variable hand speed therefore creates local distortion, not just one correctable scale factor.

For stop-and-capture operation, measure each position, wait for settling, and acquire while stationary. For continuous constant-speed operation, exclude acceleration and deceleration regions, verify actual travel over time, and retain acquisition timestamps.

An encoder on a drive shaft measures shaft motion. It represents object travel only while the mechanical relationship remains valid. If the part can slip or move independently, measure motion closer to the part or provide a validated constraint.

6. Reconstruct with position metadata intact

Store a record for every profile containing its identifier, exposure timestamp when available, travel position, validity information and calibrated profile coordinates. Preserve the original records before resampling or filtering.

For the simple case of parallel profiles perpendicular to the travel direction, the measured displacement can define the scan coordinate:

Choose the sign to match the reconstruction frame. Moving a sensor over a stationary object and moving an object under a stationary sensor do not necessarily use the same transform direction.

For a more general arrangement, transform each calibrated profile into the chosen reconstruction frame using its corresponding pose:

The rotation and translation must map the profile frame into the reconstruction frame at acquisition. Travel along one axis is insufficient when the object also rotates, lifts or rocks.

Keep irregularly spaced profiles at their measured locations. If downstream software requires a regular grid, resample afterward and retain a mask for unsupported regions. Multiple profiles captured during a pause belong at nearly the same position; they do not represent additional travel.

Avoid using smoothing to conceal missing coverage. The guide to point-cloud downsampling distortion explains why processing can change geometry even after acquisition is correct.

7. Verify dimensions before applying a scale correction

Use a reference artifact with known dimensions and uncertainty appropriate to the inspection. Measure across the laser profile and along the scan axis separately. One overall bounding-box dimension is not enough to distinguish calibration error from incorrect travel assignment.

Perform these checks before accepting the reconstruction:

• Scan the same artifact at several operating speeds. Dimensions should remain within the defined acceptance limits.
• Measure several intervals along the travel direction. A correct total length can hide alternating stretched and compressed regions.
• Inspect profile identifiers and position increments for missing records, duplicates and reversals.
• Compare repeated scans at the same speed to assess repeatability separately from dimensional bias.
• Compare flat surfaces and sharp transitions to distinguish motion effects from stripe-extraction problems.

For a known interval of 80.00 mm reconstructed as 82.00 mm:

A scale correction may be justified if that error is repeatable across positions, speeds and reference lengths. It is not justified when only one section is stretched or the error changes with speed.

If heights are unstable while scan-axis distances remain consistent, investigate laser-line extraction noise. If surface coverage changes with object height, check 3D camera working distance and positioning instead of changing the travel scale.

8. Choose hardware around the verified method

Start with the required pitch, maximum speed, measurement width, height range and tolerance. Then review 3D sensors and the camera finder for the appropriate acquisition architecture.

The MIS-DP3850-01P is listed as a laser profiler. Confirm encoder support, trigger behavior, position metadata and acquisition limits before selecting it for synchronized scanning.

The MV-DP4180-03P is also listed as a laser profiler. Request the same integration information rather than assuming these functions from the product category.

A separate camera and laser can support a development setup, but you must provide stripe extraction, geometric calibration and motion association. The LS-CMR-0166 is an area scan option with a rolling shutter. Evaluate that shutter behavior against the acquisition method before considering it for moving-profile capture.

For camera-based development, the camera configurator helps organize component selection. Obtain software, firmware and integration documentation through authorized manufacturer or distributor channels. Operate laser equipment under its documented safety requirements.

9. Complete the acceptance checklist

Before ordering or releasing the measurement process, confirm that:

• Every profile has a measured position or a justified motion-based position estimate.
• Encoder scale and direction have been checked against physical travel.
• Peak speed does not exceed the verified acquisition and processing capacity.
• Exposure movement and timing uncertainty fit the application error budget.
• Missing profiles retain their gaps instead of shifting later data.
• Coordinate transforms account for the actual sensor and object motion.
• Local intervals and total length pass dimensional checks at several speeds.
• Raw profiles and acquisition metadata remain available for diagnosis.

Equal spacing is the wrong answer when motion is irregular and unmeasured. A fixed spacing correction is also wrong when the part slips, rotates or changes speed unpredictably. In those cases, improve position measurement before adjusting the point cloud.

Take your required pitch, peak speed, reference measurements and trigger requirements to application support. Those details provide a useful basis for choosing components and checking the acquisition method before committing to a complete station.