centre of rotation update on talk from 2003 y errors in centre of rotation mike avison
Post on 28-Dec-2015
215 Views
Preview:
TRANSCRIPT
Centre of Rotationupdate on talk from 2003
Y errors in centre of rotation
Mike Avison
Welcome Skylight Users
Results - by brand: NormalisedNormalised to 33cm radius
(LE collimators)
Camera Max Y deviation (mm)
Axis (avg) 5.0
E Cam 3.1
Hawkeye 1.7
Argus 4.8 (single head)
Forte A 5.2
Forte B 5.2Skylight 2.2
Results - by brand: NormalisedNormalised to 33cm radius
(LE collimators)
Camera Max X deviation (mm)
Axis (avg) 0.8
E Cam 2.3
Hawkeye 1.3
Argus 0.3 (single head)
Forte 0.7
Skylight 1.4
Head Configuration
• Normally we measure COR errors with heads at 180º
• Reasonable to assume it may be worse in less balanced configurations (90º, 102º)
How?
• In any head configuration– If projections from each head form a
contiguous set, one file is made– If not 2 files are made
• So acquire round a full 360º of rotation– 2 complete sets of 360º data, 2 files
• Use image algebra to cut and splice a simulated set of data.
Results
Configuration Xerror Yerror
180º 2.1 5.4
102º 0.6 5.6
(single head 0.7 2.2)
How much correction takes place?
• Y error is proportional to radius
• Therefore correction cannot be perfect unless correction factor for each radius
• Are Philips doing any Y correction for head sag?
How much correction takes place?
Cam/Head sagº Predicted* Measured
Yerror Y error
1/1 0.72 4.1 2.5
1/2 0.43 2.5 1.9
2/1 0.45 2.6 4.3
2/2 0.10 0.6 2.0As measured with level-meter
* radius . tan(sagº)
As measured with COR test
Y data is aligned by IRC only at start position
Start Position head 1 at 180º head 2 at 0ºResults in this discontinuity and excessive range in Y error
If the means were aligned...
mean h1
mean h2mean h1
Effect: Halving COR Y error
Conclusions
• 102º config has similar COR performance to 180º
• Confirmation: COR error would be about ½ as big if the correction procedure was slightly more sophisticated (e.g. from 5mm to 2.5mm)
Conclusions 2• It appears Philips’ correction correction program
does do some Y correction but it is probably a constant offset derived with heads at 0º and 180º respectively and at minimum radius
• Hence – It under corrects at 33cm radius– Discontinuity at 0º and 180º– Performance would be better with offset derived form
mean Y values over 360º
Acknowledgements
• Thanks to all the users who provided raw COR data in 2003 and to Alan Britten and Anil Vara who provided the Skylight raw data this year.
top related