Implementation of Adaptive Optics in Nonlinear Microscopy for Biological Samples Using Optimization Algorithms - Adaptive Optics for Biological Imaging - page 203

60

FFT

50

Intensity squared

Sobel

40

Image variance

Wavelet

30

20

10

0

FFT Intensity squared Sobel

Metric used for the relative improvement calculation

Image variance Wavelet

FIGuRE 11.6 Relative.improvement.calculated.with.the.metric.given.on.the.horizontal.axis.ater.an.optimization.
performed.by.the.metric.given.in.the.legend.

160 0.2

0.1
0.0

Before optimization

140
120

After optimization

- 0.1
- 0.2
- 0.3
- 0.4
- 0.5
- 0.6

100
80
60
40

FFT
Intensity squared
Sobel
Image variance
Wavelet

20

0 0

20 40 60 80 100

Profile in pixel

Zernike polynomials (n,m)

(a) (b)

FIGuRE 11.7 Typical.output.data.of.an.optimization.performed.on.the.image.in.Figure.5..he.proile.(a).before.
(dark.line).and.ater.(gray.line).optimization.performed.with.the.wavelet.metric..he.location.of.the.proile.is.given.
on. Figure.11.5 . by.the.line..he.histogram.(b).shows.the.mirror.Zernike.mode.amplitude.by.metrics.

the. ive. diferent. metrics. used. to. actually. perform. the. optimization.. he. height. of. each. of. the. bars. is.
very.similar.within.each.block.(or.using.the.same.inal.metric)..Although.diferent.algorithms.produce.
diferent.percentage.improvements,.the.overall.improvement.in.image.quality.was.similar.in.all.cases..
In all cases,.no.matter.which.original.metric.was.used,.the.inal.metric.value.was.the.same.for.each.met-
ric.method..here.is.no.diference.in.contrast.or.visual.sharpness..he.optimization.is.also.repeatable.
and.leads.to.the.same.mirror.shape.
In.Figure.11.7b,.the.mirror.Zernike.mode.amplitudes.are.displayed.and.no.large.diferences.can.be.
seen.ater.optimizing.with.the.diferent.metrics..he.error.bars.correspond.to.the.standard.deviation.
of.the.coeicient.ater.10.optimizations.(they.are.there.to.show.the.variation.of.values,.rather.than.the.