Understanding the Accuracy and Resolution of Non-Contact Structured Light Scanners
Scanner accuracy is not a simple answer like you would equate to a traditional measurement device such as a caliper. Because we can produce millions of points of data at a time, you have to account for precision, accuracy, and resolution. Precision is how tight your data grouping will be. It does not account for how close to the actual real value is. For example, let’s say you have an actual material thickness of 0.250”. You collect 3 measurements of 0.199”, 0.200” and 0.201”. Your device would be precise to within +-0.001”, but it is not anywhere near the actual value. How close the measurement is to the actual value is called Accuracy. In the above example, that device would not be considered very accurate for an Engineer or Machinist, being off by 0.050”. Resolution is easily described as the point density. The more data points you have, and the closer they are together, the higher the resolution.
Just like a TV or printer (measured in DPI), the higher the resolution, the better you will be able to see the fine details of your scan. Another benefit of higher resolution is that it gives you more sample data points for a given shape or surface. Reverse Engineering software uses complex algorithms to look for geometric shapes, then using an average of those points gives you a value. Whether it be the diameter of a circle, a wall thickness, the center vertices of a cylinder, or an angle, the more data points you have, the more accurate your average value will be. Higher resolutions are also useful when looking for localized imperfections. A good visual example of this would be the hood of a car. Using a CMM machine you would gather very accurate data points. However, you would usually only gather several 100 points at the most. From there you would be able to check that the hood follows the correct curve, or for any large manufacturing defects. Very useful. Now imagine that same hood was hail damaged. Only collecting a limited number of points, it would be safe to assume that you would probably miss the fact that there are tiny imperfections all over the hood. Using a 3d scanner, you would have collected several million points of data. It would be very easy with that many data points to see any small localized imperfections.
There are several things that can add deviation to a scan. One thing that can degrade the accuracy of a scan are differences in material color and texture. Suddenly changing from a smooth to matte finish, or a dark to light finish can add small variances to the way the light behaves. This can usually be overcome by utilizing an easily removable, non-destructive finish. It provides a matte white finish to provide a higher quality scan data than could normally be acquired by scanning the part’s natural finish. Another step that can degrade the scan data is the mesh alignment. Because we usually require a 360 degree scan, or a scan that is larger than the FOV (Field of View) of the scanner, it is almost always necessary to combine multiple scans together. The scanning software will look for similar geometries and features in order to align them into one seamless scan. Although this works extremely well, it will add a little deviation over a very large object, where individual scans can number in the 100s and data points can number well over 500 million.
All of this is further complicated by the fact that some 3D scanners are adjustable to accommodate scanning different sized parts. Because the amount of data points collected remain the same, the smaller the scanning field, the higher the resolution will be. It will also increase the accuracy of the scan a little. With all that said, we only use very high quality scanners to provide us with the best possible scans available for our clients. Our scanners all collect up to 2.6 million average points and 5.2 million polygons per scan. At a 200mm diagonal FOV, the point to point resolution distance average is .075mm with an accuracy of 0.0018”. At 400mm diagonal FOV, the point to point resolution distance average is 0.165mm with an accuracy of 0.0030”. At 600mm diagonal FOV, the point to point resolution distance average is 0.250mm with an accuracy of 0.0041”. We also have a macro scanner that has a very small FOV of only 50mm. It has a point to point resolution distance average of only 0.019mm and an accuracy of 0.0006”! Obviously this will provide a very small and detailed scan of a part measuring only a few inches across.
We hope that this little explanation will give you a little better understanding into the accuracy and resolution of white light scanners. We do our best to strive for the highest accuracy and the maximum needed resolution for any particular project. It is up to the experience of the scanner’s operator to find the best combination of scanner FOV and the number of scans required for a complete 360 degree scan