subjective image quality metrics from the wave...
Post on 07-Nov-2018
217 Views
Preview:
TRANSCRIPT
Subjective Image Quality Metrics fromThe Wave Aberration
David R. WilliamsWilliam G. Allyn Professor of Medical Optics
Center For Visual ScienceUniversity of Rochester
Commercial Relationship:Bausch and Lomb
Funding:Bausch and Lomb
NSF Science and Technology Center for Adaptive Optics
Collaborators:Univ of Rochester:Li ChenNathan DobleHeidi Hofer
Ian Cox, Bausch and LombRay Applegate, University of HoustonLarry Thibos, Indiana University
Scott MacRaeJason PorterBen SingerRemy Tumbar
Goal:
To compute a number that captures the subjective effect of the eye’s wave aberration.
Uses:
Assessing severity of the wave aberrationCalculating the best correction
Defocusrms = 0.5 µm
Spherical Aberration
rms = 0.16 µm
Defocus andSpherical Aberration
rms = 0.52 µm
Some Aberrations Interact Strongly in Image Blur
Zernike Modes
trefoil coma coma trefoil
quadrafoil secondary spherical
2nd
3rd
4th
5th
radialorder
secondary quadrafoil
pentafoil secondary secondarysecondary pentafoilsecondary
defocusastigmatism astigmatism
astigmatism astigmatism
trefoiltrefoil comacoma
Z20Z 2
-2 Z22
Z 3-1 Z3
1Z 3-3 Z3
3
Z40 Z4
2Z 4-2 Z4
4Z 4-4
Z51Z 5
-1 Z53Z 5
-3 Z55Z 5
-5
Lower OrderAberrations
Higher OrderAberrations
Zernike Modes
trefoil coma coma trefoil
quadrafoil secondary spherical
2nd
3rd
4th
5th
radialorder
secondary quadrafoil
pentafoil secondary secondarysecondary pentafoilsecondary
defocusastigmatism astigmatism
astigmatism astigmatism
trefoiltrefoil comacoma
Z20Z 2
-2 Z22
Z 3-1 Z3
1Z 3-3 Z3
3
Z40 Z4
2Z 4-2 Z4
4Z 4-4
Z51Z 5
-1 Z53Z 5
-3 Z55Z 5
-5
Lower OrderAberrations
Higher OrderAberrations
Zernike Modes
trefoil coma coma trefoil
quadrafoil secondary spherical
2nd
3rd
4th
5th
radialorder
secondary quadrafoil
pentafoil secondary secondarysecondary pentafoilsecondary
defocusastigmatism astigmatism
astigmatism astigmatism
trefoiltrefoil comacoma
Z20Z 2
-2 Z22
Z 3-1 Z3
1Z 3-3 Z3
3
Z40 Z4
2Z 4-2 Z4
4Z 4-4
Z51Z 5
-1 Z53Z 5
-3 Z55Z 5
-5
Lower Order Aberrations
Higher OrderAberrations
Zernike Modes
trefoil coma coma trefoil
quadrafoil secondary spherical
2nd
3rd
4th
5th
radialorder
secondary quadrafoil
pentafoil secondary secondarysecondary pentafoilsecondary
defocusastigmatism astigmatism
astigmatism astigmatism
trefoiltrefoil comacoma
Z20Z 2
-2 Z22
Z 3-1 Z3
1Z 3-3 Z3
3
Z40 Z4
2Z 4-2 Z4
4Z 4-4
Z51Z 5
-1 Z53Z 5
-3 Z55Z 5
-5
Lower OrderAberrations
Higher OrderAberrations
There are strong interactions between Zernike Modes.
Therefore,Decomposing the wave aberration
into Zernike modes is not the best way to evaluate the subjective impact of
the wave aberration
Point Spread Function
How Bad is This Wave Aberration?
Wave Aberration
Use Retinal Image Quality, Not the Wave Aberration
Aberrations in Lens and Cornea Distort Wave front
Principle of Adaptive Optics
Deformable MirrorCorrects Wave front
Sharp Image in Camera
Wave front SensorMeasures Wave front
Adaptive Optics Can Create Wave Aberrations
Wave Aberration After AO correction With coma added
(Subject: ND �)
Convoluted retinal image
Wave aberrations from Lasik postOp patient
Wave aberrations created by adaptive optics(With real eye, JP)
Blur Matching
• Lots of Sharp Edges• Edges At All Orientations• Seen Through Adaptive
Optics
550 nm, 1 deg, 6 mm pupil
Binary Noise Stimulus
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
DefocusPatient wave aberration Stimulus
The subject adjusted the amplitude of defocus so that thesubjective blur matched that of the patient wave aberration.
Blur Matching of Patient Wave Aberrations
Zernike mode Zernike mode
Am
plitu
de(µ
m)
Def
ocus
(D)
Blur Matching Controls for Neural Differences between
Patients
Using Multiple Subjects Controls for Neural Adaptation
Acuity does not always capture thesubjective quality of vision
Equal Acuity But Different Subjective Image Quality
Compare Subject Matches with Matches Made Using Three Different Metrics
Wavefront RMSStrehl Ratio
Neural Sharpness
Wave Aberration RMS
-3 -2 -1 0 1 2 3-1
0
1
2
3
Am
plitu
de (µ
m)
Aperture(mm)
Big RMS
Small RMS
Big RMS Small RMS
R2 = 0.4627
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
RMS Metric
Met
ric m
atch
es (D
)
Subject Matches (D)
Strehl Ratio of Point Spread Function
Diffraction-limited PSF(Perfect Eye)
Hdl
Heye
Actual PSF(Aberrated Eye)
Strehl Ratio = H eye
H dl
C. of Austin Roorda
R2 = 0.4902
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
Strehl Ratio MetricM
etri
c mat
ches
(D)
Subject Matches (D)
( ) SubjectiveImage
Quality Point Spread
FunctionGaussian Neural Blur
σ = 0.8’
Σ x =
A Simple, Biologically-Plausible Metric for Subjective Image Quality
R2 = 0.7034
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
Neural Sharpness Metric
Met
ric m
atch
es (D
)
Experiment (D)
R2 = 0.7034
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
Neural Sharpness
Met
ric m
atch
es (D
)R2 = 0.4627
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
Met
ric M
atch
es (D
)
Subject Matches (D)
Wavefront RMS
Collaboration to Identify the Optimum Image Quality Metric
Ray Applegate, University of Houston:Effectiveness of Image Quality Metrics in PredictingVisual Acuity with Convolution Simulations
David Williams, University of RochesterEffectiveness of Image Metrics in predictingSubjective Image Quality with Adaptive Optics
Larry Thibos, Indiana University:Effectiveness of Image Quality Metricsin Predicting Visual Acuity in the Population
defocus
met
ric
valu
e
maximum
Optimizing refraction with an image quality metric
error
subjective
search in3-D space
Guirao and Williams (2003)
Conclusions• Generating blur with adaptive optics leads to a robust metric for correcting vision.
• It is hard to estimate subjective blur from the wave aberration. Zernike Decomposition doesn’t help much.
• The point spread function combined with a biologically plausible model of neural blur is better.
• Standards for optimizing correction from wavefront are in the works
top related