Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Chapter 1 – Math Toolbox
Gabriel Popescu
University of Illinois at Urbana‐Champaign y
p g
Beckman Institute
Quantitative Light Imaging Laboratory
Quantitative
Light Imaging Laboratory
http://light.ece.uiuc.edu
Principles of Optical Imaging
Electrical and Computer Engineering, UIUC
ECE 460 – Optical Imaging
Objectives
Develop a set of tools useful throughout the course
p
g
Chapter 1: Math Toolbox
2
ECE 460 – Optical Imaging
1.1 Linear Systems
1.1 Linear Systems
Consider a simple system:
p y
Equation of motion:
d 2x
dx
m 2 m mo 2 x f (t )
dt
dt
k
m
(1.1)
o
Define Operator: (linear differential eqs)
(
)
Lm
d
d
2
m
m
o
dt 2
dt
L( x) F (t )
Chapter 1: Math Toolbox
Mass
on a
spring
k
m
(1.2)
(1.3)
3
ECE 460 – Optical Imaging
1.1 Linear Systems
1.1 Linear Systems
Operator L has important properties:
p
p
p p
2
d (ax)
d (ax)
2
L
(
ax
)
m
m
m
(ax)
a)
o
2
dt 2
dt
d ( x)
d ( x)
a m 2 m
mo 2 ( x)
dt
dt
aL( x)
(1.4)
d ( x y)
d ( x y)
m
mo 2 ( x y )
b) L( x y ) m
dt
dt
L( x) L( y )
(1.5)
Chapter 1: Math Toolbox
4
ECE 460 – Optical Imaging
1.1 Linear Systems
1.1 Linear Systems
Definition: An operator obeying properties L(ax) = aL(x) and p
y gp p
( )
( )
L(x+y)=L(x)+L(y) is called linear
Most of the system in nature are linear; well, at least to the fi t
first approximation
i ti
They are mathematically tractable analytic solutions
Consider equations:
Consider equations:
L(x1) = 0
(1.6)
L(x
( 2)) = 0
x1, x2 are solutions
Chapter 1: Math Toolbox
5
ECE 460 – Optical Imaging
1.1 Linear Systems
1.1 Linear Systems
Continuing:
g
L(ax1+ bx2) = L(ax1) + L(bx2)
= aL(x1) + bL(x2)
(1.7)
= 0 + 0
Any linear combination of solutions: x1, x2 is also a
solution
The number of independent solutions = degrees of freedom
X 1 , X 2 ,,...,, X N = independent solutions if
independent solutions if
X i j x j , for any
(1.8)
j
j i
Linear Differential eqs of order N allow for N independent solutions
Chapter 1: Math Toolbox
6
ECE 460 – Optical Imaging
1.2 Light‐matter
1.2 Light
matter interaction
interaction
Classic model of atom: e‐ rotating around N ≈ planets
y
+
_
m
e, m
x(t)
x
x x 0 cos(0t )
0
x
t
Lorentz Model
Analogy :
w
x
Chapter 1: Math Toolbox
x x 0 cos(0t )
o
k
m
7
ECE 460 – Optical Imaging
1.2 Light‐matter
1.2 Light
matter interaction
interaction
So, motion of charge follows the same eq (1.1)
,
g
q( )
d 2x
dx
m 2 m mo 2 x F (t )
dt
dt
(1.9)
Incident field drives the charge: F (t ) qE (t )
For e-, q = -e !
Monochromatic field: E (t ) Eoeit
(1.10)
mx mx mo 2 qEoeit
This is the eq of motion for eletric charge under incident EM This is the eq of motion for eletric charge under incident EM
field. Can explain most of Optics!
Chapter 1: Math Toolbox
8
ECE 460 – Optical Imaging
1.3 Superposition principle
1.3 Superposition principle
Suppose we have 2 fields simultaneously interacting with the pp
y
g
material (Eg. ω1, ω2):
E1
E2
E1 E1 e i t ;;qE
qE1 F1
1
E2 E2 e
i2t
; qE2 F2
(1.11)
Let x
Let x1, xx2 be solutions of displacements for the two forces F
be solutions of displacements for the two forces F1
and F2
L(x
( 1)) = F1((t))
(1 12)
(1.12)
L(x2) = F2(t)
Chapter 1: Math Toolbox
9
ECE 460 – Optical Imaging
1.3 Superposition principle
1.3 Superposition principle
Consider the same solution:
L( x1 x2 ) L( x1 ) L( x2 )
F1 (t ) F2 (t )
(1 13)
(1.13)
So, final solution is just the sum of individual solutions. Nice!
E1
This is the superposition principle
E
For the 2 frequency example:
E2
E +E
1 2
E
2
x +x
1 2
x
2
E
1
x
1
It’s as if one applies the fields one by one and sums their results
Chapter 1: Math Toolbox
demo available
10
ECE 460 – Optical Imaging
1.4 Green’ss function/impulse response
1.4 Green
function/impulse response
Let the incident field, i.e driving field, have a complicated shape
h
E(t)
arbitrary
t
i
t
E(t) can be broken down into a succession of short pulses, i.e Dirac delta functions:
(1.14)
1, t = 0
(t )
0, otherwise
E (t )
E (t ' ) (t t ' )dt '
(1.15)
Chapter 1: Math Toolbox
11
ECE 460 – Optical Imaging
1.4 Green’ss function/impulse response
1.4 Green
function/impulse response
If we know the response of the system to a short pulse, , p
y
p
, (t ) ,
the problem is solved
Let h(t) be the solution to (t )
The final solution for an arbitrary force F (t ) qE (t ) is:
x(t )
E (t ' )h(t t ' )dt '
(1.16)
This is the Green’s method of solving linear problems
h(t) = Green
Green’ss function or impulse response of the system
function or impulse response of the system
h(t) Complicated problems become easily tractable!
Chapter 1: Math Toolbox
12
ECE 460 – Optical Imaging
1.5 Fourier Transforms
1.5 Fourier Transforms
Very efficient tool for analyzing linear (and non‐linear) processes
Definition: i 2 xf x
[ f ( x)]
f ( x )e
dx
F ( f x ) f ( )
(1.17)
F is the Fourier transform of f
( , , )
(
x
,
y
,
z
)
f : ; , f must satisfy:
a) f - modulus integrable
b)) f has finite number of discontinuities in the finite
domain
c) f has no infinite discontinuities
In
I practice,
i
some off these
h
conditions
di i
are sometimes
i
relaxed
Chapter 1: Math Toolbox
demo available
13
ECE 460 – Optical Imaging
1.5 Fourier Transforms
1.5 Fourier Transforms
Inverse Fourier Transforms:
1 f ( x)
f ( )e i 2 xf df
x
f ( x)
[ ( f )] f
1
x
(1.18)
(1.19)
Meaning of F.T: reconstruct a complicated signal by summing Meaning of FT: reconstruct a complicated signal by summing
sinusoidals with proper weighting
Chapter 1: Math Toolbox
demo available
14
ECE 460 – Optical Imaging
1.5 Fourier Transforms
1.5 Fourier Transforms
Fourier transform is a linear operator:
p
[af ( x) bg ( x)]
i 2 x
[
af
(
x
)
bg
(
x
)]
e
dx
a f ( x)e i 2 x dx b g ( x)e i 2 x dx
a[ f ( x)] b[ g ( x)]
Chapter 1: Math Toolbox
(1.20)
demo available
15
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
a)) Shift Theorem: if f ( ) [ f ( x)]
{ f ( x a )} f ( )e i 2 a
(1.21)
Easy to prove using definition
Eq 1.21 suggest that a shift in one domain corresponds to a linear phase ramp in the other (Fourier) domain
linear phase ramp in the other (Fourier) domain
Chapter 1: Math Toolbox
16
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
b)) Parseval’s theorem: if [ f ( x)] f ( )
f ( x) dx
2
f ( ) 2 d
(1.22)
Conservation of total energy
Chapter 1: Math Toolbox
17
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
c)) Similarity theorem: if y
f
[ f ( x)] f ( f x ) , i.e. is the F.T of f
1
[ f (ax)]
f
|a| a
(1.23)
Theorem 1.23 provides intuitive feeling for F.T
Let’s consider
Chapter 1: Math Toolbox
18
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
c)) Similarity theorem:
y
Let’s consider:
f ( f )
x
f ( x)
Δx
Δfx
(
(Figures pageII‐8)
)
x
fx
f ( f )
x
f ( x)
2Δx
Δfx
2
x
Chapter 1: Math Toolbox
fx
19
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
! Broader functions in one domain imply narrower functions in py
the other and vice‐versa
Eg. To obtain short temporal pulses of light, one needs a broad spectrum (Ti: Saph
t
(Ti S h laser)
l
)
! Only an infinite spectrum allows for ‐function pulses
s(ω)
ω
I(t)
t
Physically Impossible
Chapter 1: Math Toolbox
20
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
Before we present the last theorems, we introduce the p
,
definitions of convolution and correlation
Let
g(x)
G( )
h(x)
H( )
Convolution of g and h:
Convolution of g and h:
gh
g ( x ' )h( x x ' )dx '
(1.24)
Correlation of g and h
g h
g ( x ' )h( x ' x)dx '
(1 25)
(1.25)
Chapter 1: Math Toolbox
demo available
21
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
v
Difference between and is h(x‐x’) vs h(x’‐x), i.e. flip vs ( )
( ),
p
non‐flip of h
Particular case:
Autocorrelation: g=h
gg
g ( x ' ) g ( x ' x)dx '
(1.26)
Exercise: Use PC to show:
x
=
(Figures page II‐9)
x
Gauss
Chapter 1: Math Toolbox
=
x Gauss = Gauss
demo available
22
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
d)) Convolution theorem:
v h ] GH
[ g
(1.27)
'
'
'
[
g
(
x
)
h
(
x
x
)
dx
] G ( ) H ( )
i.e
Convolution in one domain corresponds to a product in the other. Nice!
Multiplication is always easy to do
Multiplication is always easy to do
Recall Green’s function: h(t) = the response to a ‐function
light pulse
Chapter 1: Math Toolbox
demo available
23
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
We found (Eq 1.16):
( q
)
x(t )
E (t ' ) h(t t ' ) dt '
i.e the response to an arbitrary field E(t) is the convolution E h!
Let’s take the F.T:
x( ) E ( )h( )
(1.28)
It doesn
It doesn’tt get any simpler than this
get any simpler than this
i.e if we know the impulse response h(t), (or the Green’s
response to any function) take F.T h(ω) transfer function
field E is:
x(t ) [ E ( )h( )]
Chapter 1: Math Toolbox
(1.29)
demo available
24
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
e)) Correlation theorem:
differs from only by minus sign similar theorem:
i.e [
[ g h] GH *
g ( x ' )h( x ' x)dx ' ] G ( ) H ( )*
(1.30)
Particular case: g = h (auto correlation):
[ g g ] GG G
*
Chapter 1: Math Toolbox
2
(1.31)
25
ECE 460 – Optical Imaging
1.6 Basic Theorems with Fourier Transforms
1.6 Basic Theorems with Fourier Transforms
e)) Correlation theorem:
Eg: F.T of an auto correlation is the power spectrum
Very important for both time and space fluctuating fields:
(t )
E (t ') E (t ' t )dt = auto correlation
t
[(t )] E ( ) E * ( ) S ( ) = power spectrum
(1.32)
(Wiener–Khinchin theorem)
We
We’llll meet them again later!
meet them again later!
Chapter 1: Math Toolbox
26
ECE 460 – Optical Imaging
1.7 Differential equations and Fourier Transforms
f
Let f be a function of time:
f (t )
F ( )e it d 1 ( F )
What is What is f ?
(1.33)
f
[ F ( )eit d ]
t t
it
F ( ) [e ]d
t
t
[i F ( )]eit d
1 i F
So, f F & Chapter 1: Math Toolbox
f
t
i F
(1.34)
Very Useful!
27
ECE 460 – Optical Imaging
1.7 Differential equations and Fourier Transforms
f
Great:
[ f (t )] F ( ) Then
f (t )
[
] i F ( )
t
2 f
f
Now [ 2 ] [ ( )] ii F ( )
t
t t
2 F ( )
n f
In others words: [ n ] i n n F ( )
t
(1.34)
(1.35)
Differentiation theorem
Chapter 1: Math Toolbox
28
ECE 460 – Optical Imaging
1.7 Differential equations and Fourier Transforms
f
Why 1.35 result is important? Because linear differential y
p
equations are resolved in the frequency domain more easily
Eg: Recall our e‐ revolving around nucleus under field ill i ti E(t)
illumination E(t)
d 2 x(t )
dx(t )
2
m
m
m
x(t ) qE (t )
o
2
dt
dt
(1.36)
The solution is x(t). But we can solve for x(ω)= [ x (t )] and take in the end
1
Chapter 1: Math Toolbox
29
ECE 460 – Optical Imaging
1.7 Differential equations and Fourier Transforms
f
So,, let’s take F.T of 1.36,, using
g the differentiation
theorem:
m[ 2 x( )] i mx( ) mo 2 x( ) q
qE ( )
x( )[ m 2 i m mo 2 ] qE ( )
Since q=
q=-e:
e:
e
E ( )
x( ) 2 m
i o 2
Chapter 1: Math Toolbox
(1.37)
30
ECE 460 – Optical Imaging
1.7 Differential equations and Fourier Transforms
f
Exercise: use PC to take of Eq.
q 1.37
1
“damped”
damped oscilation,
oscilation = damping factor
! Problem solved
Chapter 1: Math Toolbox
31
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
Given the electron displacement
p
as a function of
frequency, x(ω), we can define the dipole moment:
pq
qx
p ex
(1.38)
The dipole moment is a microscopic quantity; we need a
macroscopic counterpart:
Ne 2
E
P N p 2 m2
o i
Chapter 1: Math Toolbox
(1.39)
32
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
P induced p
polarization
N concentration [m-3]
But P relates to the macroscopic response of the
material , i.e. eletric susceptibility:
P o E
o = permeability of vacuum
Finally, r 1 n 2 1
r = relative permeability
11
n = refractive index
If
, as opposed
pp
to
21
Chapter 1: Math Toolbox
31
(1.40)
(1.41)
12
22
32
13
233 , material is
33 isotropic
33
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
So,, combining
g 1.39 and 1.40:
Ne 2 / m o
2
1
2
n
2
( o ) i
(1.42)
For low-n materials, such as rarefied gases,
n 2 1 (n 1)(n 1) 2(n 1)
Ne 2
1
n 1
2m o ( 2 o 2 ) i
n' in"
Chapter 1: Math Toolbox
(1.43)
34
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
2
2
2
Ne
o
n' 1
2m o ( 2 o 2 ) 2 2
2
Ne
"
n
2m o ( 2 o 2 ) 2 2
(1.44a)
(1.44b)
n’’ = Re(n)
( ) = refractive
f
index
n” = Im(n) = absorption index
Chapter 1: Math Toolbox
35
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
Eg:
g Plane wave:
E E o e i k r ; k nk o
E Eoe
ik
Eoe
ink o r
E Eoe
o
r ( n ' in " )
n " k o r in ' k o r
absorption
e
refraction
n k o absorption coefficient
"
Chapter 1: Math Toolbox
36
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
Definition:
n’((ω)) = variation of refractive index with
frequency
= dispersion
n”(ω) = absorption line shape
n’’(ω)
n’,n’’
1
(Figure page II-16)
dn’
d >0
dω
Chapter 1: Math Toolbox
ω
0
dn’’
d
dω <0
n’(ω)
dn
dn’
dω > 0
ω
37
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
Note the line shape:
p
1
2
2
2
2
( o )
1
( o )2o
1
2
1
1
o
1
/
2
o
2
Lorentz function:
f
1
1
( )
; a = width
2
a 1 ( / a )
Chapter 1: Math Toolbox
38
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
Thus the asorption
p
line is a Lorentzian:
( ;o )
1
[ ( )] e
1
2 ;
o
1
t
e
-(t)
t
The Fourier transform of a Lorentzian is an exponential!
Chapter 1: Math Toolbox
39
ECE 460 – Optical Imaging
1.8 Refraction and absorption. Dispersion
1.8 Refraction and absorption. Dispersion
Connect to quantum mechanics:
q
2 level system:
1 E1
E E1 Eo (1 o )
0 Eo
Probability of spontaneous emission/absorption:
/ lif i
‐t/tlifetime
p(t) e
exponential decay
Linewidth is Lorentz = natural linewidth
! The model of e‐ on springs was introduced by... Lorentz
Chapter 1: Math Toolbox
40
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
Fully describe the propagation of EM fields
y
p p g
Quantify how and generate each other
Ê
Ĥ
B
I
E
t
D
II
H
j
(1.45a)
t
D
III
B 0
Chapter 1: Math Toolbox
IV
41
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
Plus material equations
q
D oE P
V
E
B o H M
VI
Definitions:
charge density
E Eletric field vectors
Magnetic field vectors j E = current density
= current density
H Magnetic field vectors
D Eletric displacement P polarization
Magnetic inductance M magnetization
B Magnetic inductance Chapter 1: Math Toolbox
42
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
Let’s combine I and II (assume no free charge: = 0, =0)
(
g , j )
Use property: ( E ) (E ) 2 E
2
0
E
0
(
E
)
E
Since Take ( EqI ) :
B
( E )
(1.46)
t
2 E ( B )
t
( o H )
t
(see next slide)
( o )
t
t
Chapter 1: Math Toolbox
43
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
E ( B )
t
( o H )
t
( o )
t
t
2E
2
t
2
2
E
2
Thus: E 2 0
t
(1.47)
Wave Equation
Chapter 1: Math Toolbox
44
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
Note: 1
2 ;
v
c
v ; c = speed of light in vacuum
n
n
o o
The wave equation describes the propagation of a atime‐
dependent field (eg. pulse)
Solution: plane wave: E Eo e i (t k r )
Solution: plane wave: k 2 n ; k = wave equation
v
c
Chapter 1: Math Toolbox
45
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
Phase of the field:
(1.48)
t k r
Note: = constant describes a surface
that moves with a certain velocity
t k r constant eq of planes k
dt kdr 0
k
dr
vp
(1.49)
dt k
wave front
The surface of constant phase is traveling with velocity:
v p phase velocity
k
Chapter 1: Math Toolbox
46
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
What is the counterpart of the wave equation in the frequency p
q
q
y
domain?
i
Well, remember
t
Upon Fourier transforming, Eq. 1.47 becomes:
1
E 2 (i.i ) E ( ) 0
v
1
2 E 2 ( 2 ) E ( ) 0
v
2
Note: k
v
2 E ( ) k 2 E ( ) 0
Chapter 1: Math Toolbox
(1.50)
47
ECE 460 – Optical Imaging
1.9 Maxwell’ss Equations
1.9 Maxwell
Equations
2 E ( ) k 2 E ( ) 0
The equation above is the “Helmholtz equation”
Describes how each frequency ω propagates
Chapter 1: Math Toolbox
48