Introduction
Scanning in environments with mirrors, glass surfaces, or shiny metal can mislead sensors and reduce the quality of maps. This guide explains how to strategize routes, navigate reflective zones, and confirm outcomes.
Before You Scan: Site Preparation
1. Walk the site and identify reflective areas
Identify all large mirrors, glass walls or partitions, polished metal surfaces, and glazed facade elements along your planned route.
Ask if you can cover mirrors (for example with paper or fabric sheets).
2. Plan your route
Plan your path so you pass reflective surfaces at an angle rather than head-on.
Identify areas with solid surrounding geometry (walls, columns, shelving) that you can keep within the field of view while passing reflective zones. These non-reflective features help SLAM maintain stable localization.
During Scanning: Movement and Technique
1. Keep solid geometry visible to the sensors
Keep non-reflective structural features visible to the sensors as you move through reflective areas. Walls, door frames, columns, and ceiling fixtures should remain in the field of view alongside any reflective surfaces.
2. Approach reflective surfaces at an angle
Walk past mirrors and glass partitions at an angle wherever the route allows.
Avoid stopping directly in front of large mirrors. If you need to pause, turn slightly so the sensors face away from the mirror plane.
3. Use loop closures around reflective zones
Perform loop closure before and after reflective areas. Enter and exit a reflective zone through the same corridor or junction, and ensure overlap with data captured in the stable, non-reflective area.
Post-Scanning Quality Checks
Review the quality map. Check the areas that correspond to reflective zones. If you see any issues caused by the reflective surfaces, plan a targeted rescan of that area with more loop closures. Note that double walls and some other inaccuracies can be fixed post processing.
Check the geometry in the processed point cloud. After processing, inspect areas around mirrors and glass surfaces for inaccurate geometry which appears behind the real surface plane. Use the Point Cloud Cleanup tool in IVION to remove these points. You can find it under Site Setup → Point Cloud Cleanup.
Apply point cloud filtering. Use High Confidence in point cloud settings. You can find that setting under Data Processing>Processing Tasks>Configure Settings in NavVis IVION Processing. this will reduce noise from low-quality returns. Low-intensity, inconsistent returns from reflections can often be reduced at this filtering stage.
FAQ
How does scanning handle mirrors, windows, and reflective surfaces?
Mirrors and other reflective surfaces (glass, polished metal, marble) cause duplicate geometry in point clouds because the LiDAR beam reflects off the surface and records a false point behind it. Cover mirrors before scanning when possible. You can use the cleaning tools in NavVis IVION to remove false points from reflective surfaces during post-processing.
What steps should I take to prepare a site for scanning in reflective environments?
Before scanning, walk the site to identify reflective areas such as mirrors and glass surfaces. Plan your route to avoid direct paths in front of these surfaces and ensure solid geometry is visible to the sensors.
How can I effectively navigate around reflective surfaces during scanning?
Approach reflective surfaces at an angle and avoid stopping directly in front of them. If you need to pause, turn slightly to keep the sensors facing away from the reflective plane.
What should I do if I encounter issues in the quality map after scanning?
Review the quality map for issues in reflective zones and plan a targeted rescan of those areas, ensuring to include more loop closures for better accuracy.
How can I clean up inaccuracies in the point cloud after processing?
Inspect the processed point cloud for inaccuracies around reflective surfaces and use the Point Cloud Cleanup tool in IVION to remove erroneous points.
Is it necessary to filter the point cloud data during processing?
Yes, applying point cloud filtering, such as using High Confidence settings, can help reduce noise from low-quality returns and improve overall data quality.