.png)
.png)
.png)
1. Hardware Performance of the Device
The performance of the LiDAR sensor, point acquisition rate, measurement range, camera configuration, IMU accuracy, and multi-sensor synchronization all affect the quality of the final point cloud.
In general, a higher point acquisition rate, stable sensor synchronization, and robust fusion between IMU and SLAM algorithms help generate more continuous and reliable scanning results. However, the specified accuracy of a device is usually measured under standard test conditions. In actual projects, the final accuracy should always be evaluated together with the site conditions and operating procedures.
2. Scanning Distance and Target Surface Materials
LiDAR measures distance by receiving laser signals reflected from target surfaces. Therefore, both the scanning distance and the reflectivity of the target material directly influence data quality.
Diffuse surfaces such as painted walls, floors, wood, concrete, and matte building materials usually provide stable point cloud results. In contrast, glass, mirrors, stainless steel, water surfaces, glossy floors, and other highly reflective or transparent materials may cause abnormal laser reflections, resulting in noise, voids, duplicated structures, or local deformation.
Users should also avoid scanning objects from an excessively short or long distance. If the scanner is too close to the target, the object may fall within the blind zone of the sensor. If the distance is too far, point density may decrease.
3. Scanning Trajectory and Loop-Closure Quality
Handheld LiDAR scanners typically rely on SLAM algorithms for real-time positioning and mapping during movement. As scanning time and travel distance increase, accumulated errors may occur. Therefore, proper trajectory planning is critical.
It is recommended to design the scanning route with effective loop closure, allowing the scanner to return to previously scanned areas. This helps the algorithm recognize the same spatial location and correct accumulated trajectory errors. For large-scale scenes, segmented loop closures can be used to avoid overly long single-pass trajectories.
In long corridors, tunnels, mines, stairwells, and other environments with limited geometric features, users should pay special attention to creating sufficient overlap and revisiting previously scanned areas.
4. Walking Speed and Scanning Posture
During scanning, the device should be held steadily. Users should avoid running, shaking the device, turning abruptly, or tilting the scanner excessively. Moving too fast may reduce overlap between consecutive frames and affect algorithm matching. Rapid rotation may also cause local point cloud misalignment.
For standard scenarios, users should walk at a steady and moderate speed. In narrow spaces, stairways, doorways, corners, and areas with dense pipes or complex structures, it is recommended to slow down and keep the scanner orientation stable. For important details, users may briefly pause to allow the device to collect more complete data coverage.
5. Richness of Scene Features
SLAM-based mapping relies on geometric and visual features in the environment for positioning. If the scene is too empty, repetitive, or lacks distinctive features, the algorithm has less information available for reliable matching, which may lead to unstable positioning or accumulated errors.
Examples of challenging scenes include large plain white walls, long straight corridors, empty halls, mines, high-rise shafts, forests, and monotonous tunnels. In such environments, users can improve feature recognition by placing temporary objects such as boxes, cones, signs, control targets, or other clearly identifiable markers.
6. Lighting Conditions, People, and Moving Objects
Although LiDAR itself does not fully depend on ambient lighting, handheld scanning systems often use camera data for colored point clouds, visual-assisted positioning, or 3D reconstruction. Therefore, lighting conditions can still affect the overall result.
Before scanning, users should turn on indoor lights and maintain stable, uniform illumination. Strong backlight, flickering lights, and dark areas should be avoided whenever possible. At the same time, moving objects such as people, vehicles, and construction machinery should be minimized during scanning. Dynamic objects may create ghosting, trailing artifacts, or additional noise.
7. RTK, Control Points, and Coordinate System Settings
If a project requires real-world geographic coordinates or higher absolute accuracy, RTK, ground control points, or total station measurements should be used for coordinate constraints.
When using RTK, users should ensure that the account, coordinate reference system, projection parameters, and fixed-solution status are correctly configured. If the coordinate system used during acquisition does not match the coordinate system used during post-processing, coordinate offsets or positioning errors may occur.
For indoor, underground, heavily obstructed, or GNSS-denied environments, RTK fixed solutions may not be stable or available. In these cases, known control points are often a more suitable option for post-processing calibration and georeferencing.
8. Initialization and Data Saving Workflow
The quality of initialization before scanning can affect subsequent mapping stability. During initialization, the device should be placed on a flat and stable surface, and users should wait until the system completes initialization before moving the scanner.
At the end of scanning, users should also follow the software instructions to stop the task properly and wait until the data is fully saved. Sudden power loss, force-closing the app, or interrupting the saving process may result in incomplete data, which can affect post-processing and final point cloud generation.
9. Post-Processing Software and Computer Performance
Point cloud data usually needs to be processed through post-processing software for calculation, fusion, denoising, colorization, coordinate transformation, control point calibration, and result export. Software version, processing parameters, computer configuration, and data integrity can all affect the final result.
For large-scale scenes or long-duration scanning projects, it is recommended to use a computer with sufficient performance for processing. Users should also copy the data to a local drive before processing, instead of processing directly from removable storage, to avoid failures or low efficiency caused by insufficient read speed, memory, or GPU resources.
How to Improve Handheld LiDAR Scanning Accuracy
- Plan the scanning route in advance and create effective loop closures whenever possible
- Keep the device stable during initialization and wait until initialization is completed before moving
- Maintain a steady walking speed and avoid rapid rotation or excessive shaking
- Slow down when passing through doorways, corners, stairways, and narrow spaces
- Avoid glass, mirrors, highly reflective metals, water surfaces, and other challenging materials where possible
- Add temporary reference objects in long corridors, mines, open spaces, or other feature-poor environments
- Use RTK, control points, or total station data when geographic coordinates or higher absolute accuracy are required
- Properly save the scan data after acquisition and use the official software for post-processing
Conclusion
The accuracy of handheld LiDAR scanning is jointly influenced by device performance, site conditions, scanning trajectory, operating habits, positioning method, and post-processing workflow. Standardized data acquisition and proper project planning can effectively reduce accumulated errors, drift, and noise, improving the completeness and reliability of the final point cloud.
In practical projects, users should select an appropriate scanning strategy based on the characteristics of the site and verify accuracy with RTK, control points, or professional surveying instruments when required. This helps ensure that the final 3D data meets the intended project and delivery requirements.
