analysis of the folding processweb.mit.edu/2.717/www/2.717-wk13-a.pdf · 05/09/05 – wk13-b-13...

68
MIT 2.717 05/09/05 – wk13-b-1 3D Optics

Upload: others

Post on 23-Sep-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

  • MIT 2.71705/09/05 – wk13-b-1

    3D Optics

  • MIT 2.71705/09/05 – wk13-b-2

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-3

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-4

    Imaging a 2½D object

    2f 2fx ′′ x′x

    1f 1f

    DepthDepthOf Of

    FocusFocus

    DefocusDefocus

    zworsensworsens

    awayawayfromfromfocalfocalplaneplane

    22½½DDobjectobject

  • MIT 2.71705/09/05 – wk13-b-5

    Volume holographic aperture filter (VH lens)

    2f 2fx ′′ x′x

    1f 1f

    The MIT solution The MIT solution ((SinhaSinha et al, 2002et al, 2002--2004)2004)

    add volume hologram in the Fourier plane

    z

    in focus portionin focus portiononlyonly

    is visible!add volume hologram in the Fourier plane

    is visible!

  • MIT 2.71705/09/05 – wk13-b-6

    Shape recovery (“profilometry”) with VH lenses

    Line scan method: 2D scanningLine scan method: 2D scanninglongitudinal + one lateral dimensionlongitudinal + one lateral dimension

    raw imageson CCD camera

    2½D shape(“profile”)

    in digital form

    slitslit--likelikefield of viewfield of view

    scanning recoversscanning recoversthe full fieldthe full field

  • MIT 2.71705/09/05 – wk13-b-7

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-8

    The “4F system” (telescope) : a general model for imaging

    2f 2fx ′′ x′x

    1f 1f

    z

    imageimageplane

    objectobjectplane

    FourierFourierplane planeplane plane

  • MIT 2.71705/09/05 – wk13-b-9

    Fourier filtering

    2f 2fx ′′ x′x

    1f 1f

    z

    imageimageplane

    objectobjectplane

    FourierFourierfilter planeplane filter

    PSF)(filter (in)(out) 2D∗=

  • MIT 2.71705/09/05 – wk13-b-10

    3D aperture filtering

    2f 2fx ′′ x′x

    1f 1f

    z

    3D3D FourierFourierfilterfilter

    surf3D PSF)(filter (in)(out) ∗=

  • MIT 2.71705/09/05 – wk13-b-11

    What is a 3D pupil?

    z ′′

    y ′′

    x ′′ Special case:•• modulation on a modulation on a spatial carrierspatial carrier ::

    ( ) ( )rr rK ′′=′′ ′′⋅ εε ~e ci

    ““volume hologramvolume hologram””•• 3D spatial heterodyning3D spatial heterodyning•• optically recorded oroptically recorded orfabricated (e.g., twofabricated (e.g., two--photon)photon)

    More special case:•• modulation modulation sampledsampled in in zz direction :

    ( )r ′′ε

    3D modulation of3D modulation ofrefractive index

    direction :refractive index““multimulti--layered diffractive elementlayered diffractive element””•• 3D spatial heterodyning3D spatial heterodyning•• fabricated lithographicallyfabricated lithographically(layer(layer--byby--layer or layer or origamiorigamiTMTM))

  • MIT 2.71705/09/05 – wk13-b-12

    How volume holographic 3D pupils are madeAlternative fabrication methods

    exist, e.g. by Ondax, Inc. (proprietary)

    Optical recordingOptical recording

    two mutually coherenttwo mutually coherentlight beams (e.g. from thelight beams (e.g. from the

    same laser) interferesame laser) interfereinside 3D holographic inside 3D holographic

    materialmaterial

    focal focal pointpoint

    depth depth reference reference

    planeplane

    objectiveobjectivelenslens

    signal beamsignal beamtypically a typically a ““plane waveplane wave””

    (pencil of light)(pencil of light)

    reference beamreference beam

    has has focal pointfocal point, which, whichdetermines determines ““depth depth reference planereference plane””

  • MIT 2.71705/09/05 – wk13-b-13

    Volume holography background

    • Invented by van Heerden in 1963

    • Popular in the 70s (due to improved understanding of materials such as photorefractives and photochromics) and 90s (due to advances in optoelectronics, e.g. CCD cameras, spatial light modulators, and optomechanics)

    • Main applications: holographic data storage (promise of storage density much higher than optical disks due to the 3D nature of volume holograms), optical interconnects for high-speed computing

    • Use of volume holograms in capacity of “unusual” lenses first reported by Barbastathis, Balberg & Brady in 1999.

  • MIT 2.71705/09/05 – wk13-b-14

    Optical filtering using a 3D pupil?

    ( )yxp ,

    ( )yxq ′′,

    ( )zyx ′′′′′′ ,,ε

    ( )yxp ,Given input field

    ( )yxq ′′,seek output field

  • MIT 2.71705/09/05 – wk13-b-15

    Optical filtering using a 2D pupil (shift-invariant)

    ( )yxp ,

    ( )yxq ′′,

    ( )yx ′′′′ ,ε

    ( )yx ′′′′ ,ε

    ( ) ( )vfufvuH 11 ,, λλε=

    ( ) ( ){ }⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′′ℑ=′′22

    ,,,fy

    fxvuHyxh

    λλ

    Pupil function (2D)

    Amplitude Transfer Function

    Point Spread Function

  • MIT 2.71705/09/05 – wk13-b-16

    Optical filtering using a 3D pupil /1

    ( )yxp ,

    ( )yxq ′′,

    ( )zyxP ′′′′′′ ,,

    Field propagating in the vicinity of the Fourier plane:

    ( ) ( ) ( )⎭⎬⎫

    ⎩⎨⎧ ′′+

    −′′+′′

    ⎭⎬⎫

    ⎩⎨⎧=′′′′′′ ∫∫ 2

    1

    22

    1

    2exp,dd2exp,,f

    zyxif

    yyxxiyxypxzizyxPλ

    πλ

    πλ

    π

    (equivalent to Fourier transform of p(x,y) plus defocus term)

  • MIT 2.71705/09/05 – wk13-b-17

    Optical filtering using a 3D pupil /2

    ( )yxp ,

    ( )yxq ′′,

    ( )zyx ′′′′′′ ,,ε

    Effect of the 3D pupil: generates secondary coherent radiation of the form

    ( ) ( ) ( )zyxPzyxzyxg ′′′′′′×′′′′′′=′′′′′′ ,,,,,, ε

    assuming 1st-order Born approximation (weak hologram) is valid:• no probe depletion• no rediffraction

  • MIT 2.71705/09/05 – wk13-b-18

    Optical filtering using a 3D pupil /3

    ( )yxp ,

    ( )yxq ′′,

    ( )zyxg ′′′′′′ ,,

    Field at the output plane: Fourier transform of the 3D scattered field

    ( ) ⎥⎦

    ⎤⎢⎣

    ⎡⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′+′−

    ′′=′′ 2

    2

    22

    22 211,,,

    fyx

    fy

    fxGyxq

    λλλ

    (but output is measured on a plane, hence the Fourier transform is sliced sliced )

  • MIT 2.71705/09/05 – wk13-b-19

    Shape of the slice

    u

    v

    w

  • MIT 2.71705/09/05 – wk13-b-20

    Summary: 2D pupil vsvs 3D pupil / filtering view-point

    ( )yxp ,

    ( )yxq ′′,

    ( )zyx ′′′′′′ ,,ε

    2D2D 3D3D( ) ( )

    ( )⎭⎬⎫

    ⎩⎨⎧ ′′+−×

    ×⎭⎬⎫

    ⎩⎨⎧ ′′+′′

    ⎭⎬⎫

    ⎩⎨⎧=′′′′′′ ∫∫

    21

    22

    1

    exp

    2exp,dd2exp,,

    fzyxi

    fyyxxiyxypxzizyxP

    λπ

    λπ

    λπ( ) ( )

    ⎭⎬⎫

    ⎩⎨⎧ ′′+′′

    ⎭⎬⎫

    ⎩⎨⎧=′′′′′′ ∫∫

    1

    2exp,dd2exp,,f

    yyxxiyxypxzizyxPλ

    πλ

    π

    ( ) ( ) ( )zyxPzyxzyxg ′′′′′′×′′′′′′=′′′′′′ ,,,,,, ε( ) ( ) ( )yxPyxyxg ′′′′×′′′′=′′′′ ,,, ε

    ( ) ⎥⎦

    ⎤⎢⎣

    ⎡⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′+′−

    ′′=′′ 2

    2

    22

    22 211,,,

    fyx

    fy

    fxGyxq

    λλλ( ) ⎥

    ⎤⎢⎣

    ⎡ ′′=′′

    22

    ,,fy

    fxGyxq

    λλ

  • MIT 2.71705/09/05 – wk13-b-21

    Summary: 2D pupil vsvs 3D pupil / ATF & PSF view-point

    ( )yxp ,

    ( )yxq ′′,

    ( )zyx ′′′′′′ ,,ε

    2D2D 3D3D

    ( ) { } ⎥⎦

    ⎤⎢⎣

    ⎡⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′+⎟⎟

    ⎞⎜⎜⎝

    ⎛ ′+ℑ=′′

    2121

    1,1,;,fy

    fx

    fx

    fxyxyxh

    λλε

    Point-Spread Function(shift invariant)

    Amplitude Transfer Function

    ( ) { }

    ⎥⎥⎥⎥⎥

    ⎢⎢⎢⎢⎢

    ⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′+′−

    +

    ⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′+⎟⎟

    ⎞⎜⎜⎝

    ⎛ ′+

    ℑ=′′

    22

    22

    21

    22

    2121

    221

    ,1,1

    ,;,

    fyx

    fyx

    fy

    fx

    fx

    fx

    yxyxh

    λ

    λλε

    Point-Spread Function(shift variant)

    ( ) ( )vfufvuH 11 ,, λλε=

  • MIT 2.71705/09/05 – wk13-b-22

    Summary: 2D pupil vsvs 3D pupil / ATF & PSF view-point

    ( )yxp ,

    ( )yxq ′′,

    ( )zyx ′′′′′′ ,,ε

    2D2D 3D3D( ) ( )vfufvuH 11 ,, λλε=

    ( ) { } ⎥⎦

    ⎤⎢⎣

    ⎡⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′+⎟⎟

    ⎞⎜⎜⎝

    ⎛ ′+ℑ=′′

    2121

    1,1,;,fy

    fx

    fx

    fxyxyxh

    λλε

    Point-Spread Function(shift invariant)

    Amplitude Transfer Function

    Point-Spread Function(shift variant)

    ( ) ( ) zzf

    yxizvfufyxvuH ′′⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛′′

    ′+′′′=′′ ∫ dexp,,,;, 22

    22

    11 λπλλε

    ( ) { } ⎥⎦

    ⎤⎢⎣

    ⎡⎟⎟⎠

    ⎞⎜⎜⎝

    ⎛ ′+⎟⎟

    ⎞⎜⎜⎝

    ⎛ ′+ℑ=′′

    2121

    1,1,;,fy

    fx

    fx

    fxyxyxh

    λλε

    Amplitude Transfer Function(position dependent)

  • MIT 2.71705/09/05 – wk13-b-23

    Shift-variant imaging using 3D pupils

    Bragg mismatched:Bragg mismatched:slit effectslit effect

    x

    x′

    @Bragg match:@Bragg match:apodizerapodizer offoff--axisaxis

    PlanePlane--toto--plane wave hologram, offplane wave hologram, off--axis reference onaxis reference on--axis signalaxis signal

  • MIT 2.71705/09/05 – wk13-b-24

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-25

    Explanation of shift-variance: phase matching

    Recording Bragg-matched readout

    Diffracted beams from elemental thin slicesare all in phase in phase → strong diffraction

  • MIT 2.71705/09/05 – wk13-b-26

    Explanation of shift-variance: phase matching

    Recording Bragg-mismatched readout

    Diffracted beams from elemental thin slicesare out of phase out of phase → no diffraction

  • MIT 2.71705/09/05 – wk13-b-27

    Explanation of shift-variance: phase matching

    Recording Bragg-mismatched readout

    Rθ∆

    θ

    Λ

    L

    θ

    θλθsin2R LL

    =∆Angle Bragg selectivity

  • MIT 2.71705/09/05 – wk13-b-28

    Explanation of shift-variance: k-sphere

    Recording Bragg-matched readout

    Rk

    Sk

    Kλπ2

    radius

    =k

    pk

    K

    RS kkK −=

  • MIT 2.71705/09/05 – wk13-b-29

    Explanation of shift-variance: k-sphere

    Recording Bragg-matched readout

    Rk

    Sk

    Kλπ2

    radius

    =k

    pk

    K

    dk

    Kkk += RdRS kkK −=

  • MIT 2.71705/09/05 – wk13-b-30

    Explanation of shift-variance: k-sphere

    Recording Bragg-mismatched readout

    Rk

    Sk

    Kλπ2

    radius

    =k

    pk

    K

    dk

    zkd∆

    ( )[ ] zxxKkk ˆˆ ˆ dRd zk+⋅+=RS kkK −=k=d k

    ⎟⎠⎞

    ⎜⎝⎛ ∆= LkII zd

    2matched-Braggdiffracted 2

    1sinc

  • MIT 2.71705/09/05 – wk13-b-31

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-32

    Defocus with 3D pupils

    x′ x′x′

    y′

    λ640 =z λ2000 =z λ4000 =z

    λλ =probe

    PlanePlane--toto--plane wave hologram, offplane wave hologram, off--axis reference onaxis reference on--axis signalaxis signal

  • MIT 2.71705/09/05 – wk13-b-33

    Dispersion with 3D pupils

    x′ x′x′

    y′

    λ640 =z λ2000 =z λ4000 =z

    λλ 2.1probe =

    PlanePlane--toto--plane wave hologram, offplane wave hologram, off--axis reference onaxis reference on--axis signalaxis signal

  • MIT 2.71705/09/05 – wk13-b-34

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-35

    How volume holographic lenses operate (line-scan)

    Digital readoutDigital readout

    prepre--fabricated hologramfabricated hologramin the capacity ofin the capacity of““just another lensjust another lens””in the optical trainin the optical train

    line illuminationline illumination

    FourierFourierlenslens

    signal beamsignal beamnot presentnot present

    detector planedetector plane(CCD camera)(CCD camera)

    diffracteddiffracted beam isbeam is““continuationcontinuation”” of signalof signal

    beam; inbeam; in--focus light onlyfocus light only

    reference beamreference beamreplaced byreplaced by

    22½½DDobjectobject

    inin--focus focus light onlylight only

    NOTE: the imaging processNOTE: the imaging processdoes not require the recordingdoes not require the recording

    of a hologramof a hologram (unlike traditional(unlike traditionalholographic imaging) holographic imaging)

  • MIT 2.71705/09/05 – wk13-b-36

    How volume holographic lenses operate (line-scan)

    Digital readoutDigital readout

    prepre--fabricated hologramfabricated hologramin the capacity ofin the capacity of““just another lensjust another lens””in the optical trainin the optical train

    line illuminationline illumination

    FourierFourierlenslens

    signal beamsignal beamnot presentnot present

    detector planedetector plane(CCD camera)(CCD camera)

    diffracteddiffracted beam isbeam is““continuationcontinuation”” of signalof signal

    beam; inbeam; in--focus light onlyfocus light only

    22½½DDobjectobject

    reference beamreference beamreplaced byreplaced by

    inin--focus focus light onlylight only

    NOTE: the imaging processNOTE: the imaging processdoes not require the recordingdoes not require the recording

    of a hologramof a hologram (unlike traditional(unlike traditionalholographic imaging) holographic imaging)

    laterallateralscanningscanning

  • MIT 2.71705/09/05 – wk13-b-37

    How volume holographic lenses operate (line-scan)

    Digital readoutDigital readout

    prepre--fabricated hologramfabricated hologramin the capacity ofin the capacity of““just another lensjust another lens””in the optical trainin the optical train

    line illuminationline illumination

    FourierFourierlenslens

    signal beamsignal beamnot presentnot present

    detector planedetector plane(CCD camera)(CCD camera)

    diffracteddiffracted beam isbeam is““continuationcontinuation”” of signalof signal

    beam; inbeam; in--focus light onlyfocus light only

    22½½DDobjectobject

    reference beamreference beamreplaced byreplaced by

    inin--focus focus light onlylight only

    NOTE: the imaging processNOTE: the imaging processdoes not require the recordingdoes not require the recording

    of a hologramof a hologram (unlike traditional(unlike traditional““holographic imagingholographic imaging””) )

    longitudinallongitudinalscanningscanning

  • MIT 2.71705/09/05 – wk13-b-38

    Shape recovery (“profilometry”) with VH lenses

    Line scan method: 2D scanningLine scan method: 2D scanninglongitudinal + one lateral dimensionlongitudinal + one lateral dimension

    raw imageson CCD camera

    2½D shape(“profile”)

    in digital form

    slitslit--likelikefield of viewfield of view

    scanning recoversscanning recoversthe full fieldthe full field

  • MIT 2.71705/09/05 – wk13-b-39

    Shape recovery (“profilometry”) with VH lenses

    Line scan method: 2D scanningLine scan method: 2D scanninglongitudinal + one lateral dimensionlongitudinal + one lateral dimension

    raw imageson CCD camera

    2½D shape(“profile”)

    in digital form

    Working distance d= 50cmLongitudinal resolution ∆z FWHM< 1mm

  • MIT 2.71705/09/05 – wk13-b-40

    Increasing the working distanceSinha et al, Appl. Opt. 43:1533, 2004

    • Simple objective lens • Telephoto objective optics

    L

    2r

    d

    Objective optics

    2s

    2

    s

    2

    a r

    r (EFL)

    θλ

    θλ

    LdG

    LGz PRPRFWHM ==∆

    a

    (EFL)(EFL)L

    z

    d

    a

    2s

    2

    a aθλ

    LdGz PRFWHM =∆

    Improvement = a/rImprovement = a/r

    z

    VH lensVH lens

  • MIT 2.71705/09/05 – wk13-b-41

    Increasing the working distance

    −3 −2 −1 0 1 2 30.1

    0.2

    0.3

    0.4

    0.5

    0.6

    0.7

    0.8

    0.9

    1

    Axial displacement (mm)

    Diff

    ract

    ed In

    tens

    ity (

    a.u.

    )

    Telephoto PR VHI, d=500 mmStand alone PR VHI, d=50 mm

    Longitudinal PointLongitudinal Point--Spread Function (PSF)Spread Function (PSF)

    Simple objectiveWorking distance

    d=50.2 mm.

    Telephoto objective optics Working distance

    d=500 mm; r=2 mm.

  • MIT 2.71705/09/05 – wk13-b-42

    Increasing the resolutionSinha et al, Optics Express 11:3202, 2003

    • Normal illumination • Inclined illumination

    z

    d Telephotooptics

    d

    Φ

    x∆OLDz∆

    z∆NEWz∆

    z

    Digitalcamera

    object

    Fourier lens

    VH lens

  • MIT 2.71705/09/05 – wk13-b-43

    Increasing the resolution

    NA=0.024; θs=12o; a=8 mm; d=460 mm; L=2 mm and Φ= 0o / 30o.

    Longitudinal PSFLongitudinal PSF

  • MIT 2.71705/09/05 – wk13-b-44

    Image of drug capsule using line scanCapsule provided by Pfizer

    obtained using rotary stageobtained using rotary stage

  • MIT 2.71705/09/05 – wk13-b-45

    Working distance and resolution (line scan method)Microturbine provided by

    Chee Wei Wong, Alan Epstein, MIT Telephoto + tilted illuminationTelephoto + tilted illumination

    z=0 µm z=50 µm

    z=200 µm z=250 µm

    originalobject Object Distance = 46 cmObject Distance = 46 cm

    Resolution accomplished Resolution accomplished ≤≤ 100 100 µµmmraw imagesraw images

    profileprofilereconstructionreconstruction

  • MIT 2.71705/09/05 – wk13-b-46

    NA=0.024; θs=12o; a=8 mm; d=460 mm; L=2 mm and Φ=30o.

    Image of MEMS device acquired using line scanat long working distance Nanogate provided by

    James White, Alex Slocum, MIT La

    tera

    l dim

    ensi

    on (m

    m)

    ∆z=0 µm ∆z=50 µm

    ∆z=100 µm ∆z=150 µm

    raw imagesraw images

    reconstructionreconstruction

    Original object

  • MIT 2.71705/09/05 – wk13-b-47

    Tunable grating fabricated by Wei-Chuan Shih, MIT

    Image of MEMS device acquired using line scanat long working distance

    object under microscopeobject under microscope

    ∆z=0 µm

    rawrawimagesimages

    ∆z=24 µm

    reconstructionreconstruction

    Microscope Objective NA = 0.65,

    Working distance (d) = 2 cm

  • MIT 2.71705/09/05 – wk13-b-48

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-49

    Rainbow VH imaging

    grating

    iris white light source

    objectplanecollimator

    lens

    cylindricallens

    θs

    camera

  • MIT 2.71705/09/05 – wk13-b-50

    Reduce scanning: #2 / rainbow method22½½D object is illuminated by D object is illuminated by rainbowrainbow

    (analyzed white light) (analyzed white light) ……… each matched depth is obtained simultaneously on the camera plaeach matched depth is obtained simultaneously on the camera planene

    (no slit effect (no slit effect →→ no lateral scanning necessary)no lateral scanning necessary)while preserving the outwhile preserving the out--ofof--focus light rejection propertyfocus light rejection property

    Longitudinal scanning still necessary

    Longitudinal scanning still necessary

    slice #1slice #1 slice #2slice #2

    digital reconstruction of object profiledigital reconstruction of object profile

  • MIT 2.71705/09/05 – wk13-b-51

    Imaging using the rainbow VH method

  • MIT 2.71705/09/05 – wk13-b-52

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-53

    3D→2D spatial mapping in the multiplex method

    C

    F

    I

    J

    K

    L

    Q

    R

    S

    B

    E

    H

    M

    P

    A

    D

    G

    N

    O

    T

    U

    V

    X

    W

    Y

    Z

    Θ

    camera

    A B C

    D E F

    G H I

    C

    F

    I

    CA B J

    K

    L

    JMNO P Q

    Q

    R

    S

    VHI system

    object

  • MIT 2.71705/09/05 – wk13-b-54

    Recording the multiplex VH lens

    holographicholographicmaterial

    Reference beam:Reference beam:spherical wave @ spherical wave @ zz11

    materialxx

    zz

    Signal beam:Signal beam:plane wave @ plane wave @ θθ11

  • MIT 2.71705/09/05 – wk13-b-55

    Recording the multiplex VH lens

    50µm

    zz

    xxholographicholographic

    materialmaterial

    Reference beam:Reference beam:spherical wave @ spherical wave @ zz22

    Signal beam:Signal beam:plane wave @ plane wave @ θθ22

  • MIT 2.71705/09/05 – wk13-b-56

    Recording the multiplex VH lens

    50µm

    zz

    xxholographicholographic

    materialmaterial

    Reference beam:Reference beam:spherical wave @ spherical wave @ zz33

    Signal beam:Signal beam:plane wave @ plane wave @ θθ33

  • MIT 2.71705/09/05 – wk13-b-57

    Diffraction from the multiplex VH lens

    50µm

    zz

    xxVHVHlenslens

    Fourier Fourier lenslens

    Probe beam: three point Probe beam: three point sourcssourcs@ @ zz11, , zz22, , zz3 3 (mutually (mutually incoherentincoherent))

    Reconstruction: three plane waves Reconstruction: three plane waves @ @ θθ11, , θθ22, , θθ33

    CameraCamera detects:detects:three point imagesthree point images

  • MIT 2.71705/09/05 – wk13-b-58

    Reduce scanning: #1 / multiplex methodthree three ““slicesslices”” throughthrough

    fluorescent (3D) object, stacked fluorescent (3D) object, stacked along along longitudinallongitudinal direction direction ……

    …… are viewed are viewed simultaneouslysimultaneously and and sideside--byby--sidesideon the digital cameraon the digital camera

    3D object:fluorescent beads

    in water

    implemented using three volume holographic lensesimplemented using three volume holographic lensessharing the sharing the samesame volume holographic materialvolume holographic material

    (multiplexed)(multiplexed)

  • MIT 2.71705/09/05 – wk13-b-59

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-60

    Inter-Symbol Interference (ISI) and the Viterbi algorithm

    Contributions to central pixel on raw imageContributions to central pixel on raw imagefrom voxels in object spacefrom voxels in object space

    everything else:everything else:ISIISI

    WANTEDWANTED

  • MIT 2.71705/09/05 – wk13-b-61

    Reducing complexity

    2D Viterbi algorithm with ISI size of m*n typically has statesnm2 ⋅

    Surface constraint: only one “1” on each row, number of states reduce to mn

    1)/2(mn +Use FDVA (Feedback Decision Viterbi Algorithm), number of states reduce to

    Scanning direction constraint eliminates 4/5 possible states transition paths

  • MIT 2.71705/09/05 – wk13-b-62

    Multiplex + Viterbi reconstruction

    zz11

    zz22

    11

    22

    •• 2 VHI lenses 2 VHI lenses (multiplexed inside the same crystal)(multiplexed inside the same crystal)•• 8 target reconstruction depths8 target reconstruction depths

  • MIT 2.71705/09/05 – wk13-b-63

    Multiplex + Viterbi reconstruction

    Layer #3Layer #3

    Layer #6Layer #6

  • MIT 2.71705/09/05 – wk13-b-64

    Multiplex + Viterbi reconstruction

    Layer #3Layer #3

    Layer #6Layer #6

    •• Layers #3 and #6 simultaneously in focus Layers #3 and #6 simultaneously in focus (multiplexed, imaged on different parts of the camera die)(multiplexed, imaged on different parts of the camera die)

    •• Remaining layers: reconstructed digitallyRemaining layers: reconstructed digitally

  • MIT 2.71705/09/05 – wk13-b-65

    Outline

    • Introduction• Point-spread function of 3D pupils• Phase matching (aka Bragg matching)• Defocus response of 3D pupils• Optical slicing

    – scanning– rainbow– multiplex

    • Computational approaches– maximum likelihood (Viterbi)– tomography (Radon transform)

  • MIT 2.71705/09/05 – wk13-b-66

    True 3D imaging: the Radon transform setup

  • MIT 2.71705/09/05 – wk13-b-67

    True 3D imaging: gummy bear via the Radon transform

    360 angular positions360 angular positions

  • MIT 2.71705/09/05 – wk13-b-68

    M. C. Escher“Reptiles”

    3D OpticsOutlineOutlineImaging a 2½D objectVolume holographic aperture filter (VH lens)OutlineThe “4F system” (telescope) : a general model for imagingFourier filtering3D aperture filteringWhat is a 3D pupil?How volume holographic 3D pupils are madeVolume holography backgroundOptical filtering using a 3D pupil?Optical filtering using a 2D pupil (shift-invariant)Optical filtering using a 3D pupil /1Optical filtering using a 3D pupil /2Optical filtering using a 3D pupil /3Shape of the sliceSummary: 2D pupil vs 3D pupil / filtering view-pointSummary: 2D pupil vs 3D pupil / ATF & PSF view-pointSummary: 2D pupil vs 3D pupil / ATF & PSF view-pointShift-variant imaging using 3D pupilsOutlineExplanation of shift-variance: phase matchingExplanation of shift-variance: phase matchingExplanation of shift-variance: phase matchingExplanation of shift-variance: k-sphereExplanation of shift-variance: k-sphereExplanation of shift-variance: k-sphereOutlineDefocus with 3D pupilsDispersion with 3D pupilsOutlineHow volume holographic lenses operate (line-scan)How volume holographic lenses operate (line-scan)How volume holographic lenses operate (line-scan)Increasing the working distanceIncreasing the working distanceIncreasing the resolutionIncreasing the resolutionImage of drug capsule using line scanWorking distance and resolution (line scan method)OutlineRainbow VH imagingReduce scanning: #2 / rainbow methodImaging using the rainbow VH methodOutline3D2D spatial mapping in the multiplex methodReduce scanning: #1 / multiplex methodOutlineMultiplex + Viterbi reconstructionMultiplex + Viterbi reconstructionMultiplex + Viterbi reconstructionOutlineTrue 3D imaging: the Radon transform setupTrue 3D imaging: gummy bear via the Radon transform