Testing the accuracy of ATOS[TM] 3d optical scanner measuring volumes.

Brajlih, Tomaz ; Valentan, Bogdan ; Drstevenser, Igor 等

Abstract: Non-contact optical three-dimensional measuring, scanning and digitising are increasingly present in quality assurance systems. Simple scanning procedures, high density of data acquired in a single scan, possibility of integrated reverse engineering and inspection are all advantages of optical scanning compared to conventional coordinate measuring. However, the accuracy of most optical scanners is still not up to the level of coordinate measuring machines. This paper presents the accuracy testing of three different measuring volumes of the ATOS[TM] mark II optical scanner. For this purpose, the scanned and digitized data of the test piece was compared to the measuring data collected by the ZEISS UMC 850[TM] coordinate measuring machine.

Key words: Optical scanning, 3d digitising, coordinate measuring, reverse engineering

1. INTRODUCTION

The ATOS[TM] system is based on the triangulation principle: The sensor unit projects different fringe patterns onto the object to be measured which are then recorded by two cameras (Figure 1). Each single measurement generates up to 4 million data points. In order to digitize an object completely, several individual measurements are required from different angles. Based on reference points (circular markers), which are applied to the object directly or to the measuring plate or a fixture, ATOS transforms these individual measurements fully automatically into a common global coordinate system (GOM 2007).

[FIGURE 1 OMITTED]

ATOS[TM] II three-dimensional scanner is equipped with three different projector and camera lenses setups that enable the scanning to be performed inside three different measuring volumes (135, 350 and 1200mm envelope). Using the larger measuring volume quickens the scanning of large parts by reducing the number of required consequent scans. However, the accuracy of scans falls with an increased volume envelope size. The purpose of this test was to establish the difference in dimensional accuracy of scans taken in individual volumes. The part measured was a standard demo-part for coordinate measuring machines. For a reference CAD model that was used to inspect the scanned data the demo-part was reverse-engineered directly from the measuring data acquired with ZEISS UMC 850 coordinate measuring machine (Acko, 1999). This model was later imported into ATOS[TM] 3d scanner software where comparison and inspection of the scanned data was made.

2. DIGITISING THE TEST PART

The test part was scanned inside all three measuring volumes separately. Because scanner only record points that are seen by both cameras in individual scans, it is not possible to scan the entire part with a single scan. Therefore, the whole part was digitised with several scans from different directions and subsequent scans were aligned by the reference points that were applied to the part. The preliminary result of a three-dimensional digitising is a point-cloud file that is not well suited for further analysis. Therefore, point-clouds taken in each volume were polygonised into three-dimensional polygonal models for further data processing (Pogacar 2007).

3. INSPECTING DIGITISED DATA

In order to inspect digitised data the CMM analysis between the polygonal scanned models and previously reverse engineered CAD model was made inside ATOS[TM] GOM software package. The first step is importing the reference CAD model and the registration (mutual alignment) with the scanned model (Figure 2).

[FIGURE 2 OMITTED]

The registration is made in two phases. Firstly, the approximate manual registration is made by selecting four coincident points on each model. Finer alignment is made automatically by the software with best fit registration in the second phase.

In order to verify the alignment and roughly evaluate the scanning accuracy the colour coded mesh deviation between scanned and CAD model for each measuring volume was generated (Figure 3).

[FIGURE 3 OMITTED]

For precise analysis the CMM inspection of several primitives (well defined geometrical elements) of both models was made. For the purpose of this test circle diameters from cylindrical elements and the distances between circle centres were defined as CMM elements (Figure 4).

[FIGURE 4 OMITTED]

After the definition and linking of corresponding elements between both models ATOS[TM] software automatically generates the CMM inspection report that includes data of the scanned and imported CAD element and deviation between both. Similar digitising in inspection procedure was made for each of the measuring volumes. For each volume a total of 36 individual elements were inspected.

4. RESULT ANALYSIS

For comparison of measuring volume deviations the average absolute deviations (for circle diameters and centre distances) were taken (Figure 5, 6).

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

The results show the expected difference in accuracy between measuring volumes. Somewhat surprising is the accuracy difference of circle diameter and centre distances deviations in the same measuring volume. Deviations of diameters are 2 to 3 times bigger then deviation of centre distances. This can be seen as a consequence of optical scanning limitations in digitising holes, pockets and similar features. Also, bigger measuring volumes have problem scanning smaller details due to their smaller resolution. Because the scanner software enables the combination of different measuring volumes in a single project, the future research will be conducted in optimizing various measuring volume layouts, regarding to the scanned objects properties.

5. REFERENCES

Acko, B. (1999). Industrial Measuring, Faculty of Mechanical Engineering Maribor, ISBN: 86-435-0261-8, Maribor

Pogacar, V. (2007). Integrated renaissance of design, Proceedings of 1st DAAAM Int. Spec. Conf, Drstvensek; I., Katalinic, B. (Ed.), pp. 9-12, ISBN: 3-901509-61-5, Slovenia, 4/2007, Celje

Drstvensek, I., Strojnik, T., Brajlih, T., Valentan, B. (2007). Rapid Technologies supporting surgical operation case study, Proceedings of 1st DAAAM Int. Spec. Conf, Drstvensek; I., Katalinic, B. (Ed.), pp. 53-56, ISBN: 3901509-61-5, Slovenia, 4/2007, Celje

Quality control of injection moulded parts, Available from: http://www.gom.com/EN/B0C.html Accessed: 2007-31-05