Handling noise on reflective surfaces in laser scanning

Highly reflective surfaces such as stainless steel pipes, tanks, and polished metals are a common challenge in terrestrial laser scanning. These materials often cause multipath reflections, ghost points, or noisy clusters within the point cloud. If left unaddressed, such noise can distort measurements or lead to incorrect modeling.

This article explains why reflective surfaces create noise, and how to minimize and clean up these effects, both during data capture and post-processing.

Why do reflective surfaces cause noise?

Laser scanners rely on light reflections to measure distances. When a surface is highly polished, the laser beam may scatter or bounce multiple times, leading to:

• False points (ghosting) where the scanner misinterprets the return signal.
• Noise clusters around the real surface due to irregular reflections.
• Gaps or missing data, especially on cylindrical or shiny objects like pipes.

1. Best practices during scanning

The best way to handle reflective surfaces is to prevent noise at the source:

• Use temporary matte coatings
  Apply scanning sprays (e.g., removable chalk-based sprays) or matte stickers on shiny surfaces to reduce reflections.

• Optimize scanning angles
  Avoid steep angles relative to reflective surfaces. A shallow angle reduces the intensity of direct reflections.

• Slow down the scan
  Use a higher-quality (slower) scan mode if available. Longer exposure times help the sensor pick up more reliable data.

• Control ambient light
  Strong overhead lighting or sunlight may worsen reflection issues. Dimming or diffusing lights during scanning can help.

2. Post-Processing Techniques

Even with the best scanning practices, some noise will remain. Modern point cloud software offers several filtering options to clean up data:

Apply stray point filters

Remove isolated points that are not connected to the main geometry. Most point cloud software (e.g., Scene, CloudCompare, or ReCap) has built-in stray or outlier filters.

Statistical outlier removal (SOR)

This filter checks for points that deviate significantly from their neighbors. By adjusting parameters such as search radius and minimum neighbors, you can remove noise clusters while preserving important geometry.

Clipping and manual cleanup

For stubborn noise (especially around pipes or tanks), use clipping boxes to isolate specific areas and manually delete unwanted points.

3. External tools

If the built-in tools are insufficient, export the point cloud to open-source software like CloudCompare.
CloudCompare’s “Noise Filtering → Statistical Outlier Removal” and “SOR Filter” are powerful for removing reflective-surface noise.

4. Setting expectations

• Reflective metal surfaces will always create challenges for time-of-flight and phase-based scanners. No filtering method will completely eliminate noise without losing some data.
• For highly reflective or complex geometry, consider combining terrestrial laser scans with structured-light or handheld 3D scans (using scanning spray), and merge the datasets for improved accuracy.