Systematic Errors in Dimensional X-ray Computed Tomography
Måletekniske dage
Teknologisk Institut
31.05.2012
Jochen Hiller
2 DTU Mechanical Engineering, Technical University of Denmark
Industrial X-ray CT today
Dimensional CT as a key technology in production metrology
Errors sources and a good practice in CT scanning
Conclusions and future works
Overview
3 DTU Mechanical Engineering, Technical University of Denmark
Industrial X-ray CT today
Phoenix
Zeiss
Werth Skyscan
Wenzel
4 DTU Mechanical Engineering, Technical University of Denmark
Aluminum casting
Fiber-composite
Insulin pen
Tissue Dental impression
Foam Turbine plate
Seed
Industrial X-ray CT today
5 DTU Mechanical Engineering, Technical University of Denmark
Industrial X-ray CT today
Industrial
X-ray CT
Material Analysis
Material Testing
Dimensional (geometrical) Metrology
6 DTU Mechanical Engineering, Technical University of Denmark
Dimensional CT as a key technology in production metrology
Measurement of size, form, and position CAD/CT comparison
7 DTU Mechanical Engineering, Technical University of Denmark
Dimensional CT as a key technology in production metrology
8 DTU Mechanical Engineering, Technical University of Denmark
Dimensional CT as a key technology in production metrology
We will never know the true value of a measurement
Measurement results must be repeatable and reproducible
What about systematic (effects) errors? Should be corrected!
Measurement uncertainty U:
9 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
10 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
Image artefacts
Scaling (voxel size) error
CT system limits (image blurring, noise)
Metrological data evaluation strategy
Segmentation and surface
determination errors
What are sources of systematic errors?
11 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
Only for the compensation of effects linked to geometrical scanner misalignment or beam-hardening artefacts
Calibrated masterpieces
Systematic scanning and evaluation planning to avoid high systematic errors (blunder)
Can we use calibration artefacts?
Ball-plate Ball-bar
Step-wedge
12 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
Beam-hardening Cone-beam Misalignment Undersampling
Truncation Ring artifacts Metal artifacts Noise artifacts
13 DTU Mechanical Engineering, Technical University of Denmark
X-ray tube Rotation
axis
Object
Detector
Helical sampling
z
Errors sources and good practice in CT scanning
Cone-beam artefacts:
X-ray tube
Object
Rotation
axis Detector
Circular sampling
Missing data
14 DTU Mechanical Engineering, Technical University of Denmark
Standard-CT Helical-CT
Me
asu
rem
en
t e
rro
r (µ
m)
0
5
10
15
20
25
30
35
Standard-CT Helical-CT
Me
asu
rem
en
t e
rro
r (µ
m)
0
5
10
15
20
25
Big drill hole Small drill hole
Voxel size: 105,7 µm3
Errors sources and good practice in CT scanning
Standard vs.helical CT:
15 DTU Mechanical Engineering, Technical University of Denmark
Good practice:
Tilted position of the workpiece
Errors sources and good practice in CT scanning
Sideview
Source
Detector
16 DTU Mechanical Engineering, Technical University of Denmark
Penetration length (mm)
P =
-ln
(I/I
0)
Errors sources and good practice in CT scanning
Beam-hardening:
17 DTU Mechanical Engineering, Technical University of Denmark
Penetration length (mm)
Aluminum
Iron
rela
tive
Inte
nsi
ty
Step wedge Radiography Linerization of projections with inverse function
Errors sources and good practice in CT scanning
Beam-hardening correction:
18 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
DTU beam-hardening correction GUI
19 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
Beam-hardening correction:
20 DTU Mechanical Engineering, Technical University of Denmark
uncorrected corrected 0
50
100
150
200
250
300
350
Measure
ment
err
or
(µm
)
Big drill hole
uncorrected corrected 0
50
100
150
200
Me
asu
rem
en
t e
rror
(µm
)
Small drill hole
Voxel size: (156,7 × 156,7 × 179,3) µm3
Errors sources and good practice in CT scanning
Beam-hardening correction:
21 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
Good practice:
Tilted position of the workpiece
Sideview
Source
Detector
Using a prefilter
22 DTU Mechanical Engineering, Technical University of Denmark
SRD SDD
Detector
Source Calibrated length: 8,7678 mm
v = 30 µm, Δv = 1 µm
l = 1,5 mm
Error: Δl = 50 µm
LCT
Errors sources and good practice in CT scanning
m = SDD/SRD v = a/m
23 DTU Mechanical Engineering, Technical University of Denmark
Variations are caused by
inaccuracies of the SRD
measurement (manipulator)
+
Focus drift during scanning
due to tube temperature
changes
superimposed
Errors sources and good practice in CT scanning
1 2 3 4 5 6 7 88.760
8.761
8.762
8.763
8.764
8.765
8.766
8.767
8.768
8.769
8.770
Measurement
Dis
tance (
mm
)
Reference value
24 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
Voxel size rescaling:
1 2 3 4 5 6 7 88.760
8.761
8.762
8.763
8.764
8.765
8.766
8.767
8.768
8.769
8.770
Measurement
Dis
tance (
mm
)
Reference value
25 DTU Mechanical Engineering, Technical University of Denmark
Errors sources and good practice in CT scanning
Good practice:
In particular at high magnifications: Ball-bar must be scanned together
with the workpiece
Sideview
Source
Detector
Ball-bar
Focus drift differs from scan to scan!
26 DTU Mechanical Engineering, Technical University of Denmark
Image unsharpness Noise
Object
CT
Depending on object, scanning parameters, system hard- and software
Errors sources and good practice in CT scanning
27 DTU Mechanical Engineering, Technical University of Denmark
Conclusions and future works
Importance of a consistent procedure in CT scanning planning
Variety of error sources and influence quantities in CT metrology
Possibilities to reduce systematic errors (effects)
CT as a powerful and flexible tool in production metrology
28 DTU Mechanical Engineering, Technical University of Denmark
Conclusions and future works
Material, shape, penetration lengths
Fixture, positioning, orientation
Tube voltage, current, prefilter, detector settings
Evaluation of detector images histogram analysis
Image quality (artefacts, sharpness, noise)
Voxel histogram analysis, threshold tests
Surface quality inspection
Alignment
Measurement strategy (elements, points, methods)
Reference data available, repeated measurements
Scanning preparation
29 DTU Mechanical Engineering, Technical University of Denmark
Invitation to Conference on
“Industrial Applications of CT Scanning –
Possibilities & Challenges in the Manufacturing Industry”
June 12, 2012, 10:00-16:30
DTU, Building 101, meeting room 1
2800 Kgs. Lyngby, Denmark
30 DTU Mechanical Engineering, Technical University of Denmark
Thank you very much!