Limits of the system and effective accuracy
The theoretical resolution of a 3D scanner, such as the 0.03 mm of the Revopoint MetroX, describes the smallest possible spatial distance between two detectable points in the measurement volume. This specification is typically based on the physical properties of the sensor system, such as the number of pixels of the image sensor used, the optics, the projection system and the geometry of the arrangement. In practical applications, however, the effective resolution is slightly reduced due to a number of technical influences, which are explained in more detail below. The 0.03 mm can certainly be achieved with an ideal, rectangular measuring body (caliper vs. CAD).
A central element of 3D acquisition is the generation of a point cloud. Each point on the object surface is assigned a three-dimensional coordinate in space. In structured light methods, as we have just read, this is done by evaluating light patterns that are projected onto the surface and then detected by the camera. The accuracy of this position determination depends, among other things, on the quality of the edge detection, the contrast of the projected pattern and the calibration between the projector and camera.
In practice, however, various factors lead to deviations from the idealized resolution. On the one hand, the surface quality of the object to be captured is highly relevant. Shiny, transparent or highly structured materials reflect light in unpredictable ways or scatter it, which impairs the accuracy of triangulation. Ambient light can also overlay or attenuate the projected pattern, which has a disruptive effect, particularly in the case of high ambient brightness or spectral overlay.
The signal-to-noise ratio of the optical system must also be taken into account. Even with optimal calibration, thermal noise, non-linear sensor responses or slight adjustment errors can cause the actual point position to deviate from the idealized value. These errors add up, especially for large scans with high point density requirements.
Another aspect is discretization in image processing. Edge detection and subpixel interpolation, which are necessary to determine the point coordinates with a higher accuracy than the pixel density of the sensor, are always subject to limited precision. Lens distortion correction and any rectification algorithms also contribute to the fact that the nominal resolution is not achieved equally at every point in the field of view.
Once the point cloud has been generated, it is often converted into a polygonal mesh. The discrete points are connected by a network of triangles or polygons, which necessarily means smoothing or simplification. The triangulation algorithm must decide which points to connect, making assumptions about the local geometry. Especially in areas with low point density, highly curved surfaces or undercuts, this can lead to topological artifacts. Subsequent post-processing, such as mesh optimization or gap filling, also further reduces the level of detail compared to the original point cloud.
What always helps with glossy and transparent surfaces, however, is a special spray that mattes the surface and also dissolves by itself after a few hours. The particles in the µm range are not a real source of error here, but you shouldn’t overdo it when spraying. But it’s good to know that a spray like this can minimise or even completely solve many problems.
It can therefore be said that the difference between the theoretical and practical resolution of a 3D scanner is an unavoidable consequence of optical, physical and algorithmic limitations. While the theoretical resolution serves as an idealized parameter of the system, the effective resolution describes the level of detail achievable under real conditions, which is always below the theoretically possible maximum and can vary greatly depending on the environment and object. If you know and take all this into account, you can be quite happy with the device. With a little practice and consideration of what you want to measure and how.
- 1 - Introduction, unboxing and features
- 2 - The principle of the structured light method
- 3 - The functional principle of point clouds
- 4 - Triangulation of the point clouds in a mesh
- 5 - 3D for Dummies: One-Click-Solution
- 6 - Limits of the system and the effective accuracy
- 7 - Summary and conclusion
- 8 - Appendix: Datasheet
- 9 - Appendix: Quick Start Quide
14 Antworten
Kommentar
Lade neue Kommentare
Urgestein
1
Veteran
1
Urgestein
Veteran
1
Urgestein
1
Mitglied
1
Urgestein
Urgestein
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →