Lapalician &

Homomorphic Filter

1
 ∂2 f ∂2 f
𝑄 f =
2
2
+
∂t
∂z2
Laplacian in Spatial Domain

● Laplacian
– Isotropic
– Rotation Invariant

𝑄 2 f = f ( x +1, y) + f ( x - 1, y)
+
f ( x, y +1) + f ( x, y - 1) - 4 f ( x,
y) 2
 ∂2 f ∂2 f
𝑄 f =
2
2
+
∂t
∂z2
Laplacian in Frequency Domain

3
Laplacian in Frequency
Domain
H (u, v) = - 4𝑢 2 (u 2 + v 2 )

With respect to center of frequency rectangle :

[
H (u, v) = - 4π 2 (u - M / 2)2 + (v - N /
2)2 ) ]
= - 4 π 2 D 2 (u, v)

Laplacian of an image : 4
 Laplacian in Frequency
Domain
● Enhancement Eq: g( x, y) = f ( x, y) + c [▽ 2
f ( x,

y) ]
c =- 1

● Scales of f ( x, y) & ▽ f ( x, y ) as computed by

2

DFT differ widely due to the DFT process

● Normalize f ( x, y) to [0,1] before DFT
● Normalize ▽ 2 f ( x, y) to [-1,1] 5
Laplacian in
Frequency
Domain

6
Comparative Laplacian in
Spatial & Frequency
Domains

7
Unsharp Mask, Highboost Filtering & High-
Frequency-Emphasis Filtering

● In spatial domain:
g mask ( x, y) = f ( x, y) - f ( x, y)
g( x, y) = f ( x, y) + k * g mask ( x, y)

k =1: Unsharp Masking

k > 1: Highboost
Filtering
k ∈ 1: De - emphasized
8
Unsharp Mask, Highboost Filtering & High-
Frequency-Emphasis Filtering

[ 1+ k*Hhp (u,v)] F(u,v) }

9
Unsharp Mask, Highboost Filtering & High-
Frequency-Emphasis Filtering

● In frequency domain:

g(x,y)= {[ k1+k2 * Hhp (u,v)] F(u,v) }

High- Frequency Emphasis Filter

k1>=0 : Controls the offset from origin

k2 >=0: Controls the contribution of high frequencies

10
Image:
416x596

D0=40
(5% of
short
side of
padded
image)

k1=0.5

k2=0.75
11
 Homomorphic Filtering
● Homomorphic filtering is a FDS that aims at a
simultaneous increase in contrast & dynamic range
compression.
● It is mainly utilized for non-uniformly illuminated
images in medical, sonar images etc. for edge
enhancement that makes the image details clear to the
observer.
● Certain situations where the image is subjected to the
multiplicative interference or noise as depicted
● f(x,y)= i(x,y) . r(x,y)

12
 Homomorphic Filtering…
● Illumination-Reflectance Model in FDS
● Illumination Component
Slow Spatial Variations & Attenuate contributions by illumination

● Reflectance Component
– Varies abruptly – junctions of dissimilar objects
– Amplify contributions by reflectance
● Simultaneous dynamic range compression & contrast
enhancement
–
● We
– cannot easily use the product i & r to operate

separately on the frequency components of illumination

& reflection because the FT of f ( x , y) is not separable;

1
3
 H F….
• F[f(x,y)) not equal to F[i(x, y)].F[r(x, y)].

We can separate them by taking logarithm

ln f(x,y) = ln i(x, y) + ln r(x, y).

F[ln f(x,y)} = F[ln i(x, y)} + F[ln r(x, y)]

F(x,y) = I(x,y) + R(x,y), where F, I & R are the FTs ln f(x,y),ln i(x, y) , &
ln r(x, y). respectively.

F is FT of the sum of 2 images: a low-freq illumination image (suppress) & a

high freq reflectance (enhance)image

0 < i(x,y) < a, It indicate the perfect black body

0 < r(x,y) < 1, It indicate the perfect white body

14
 H.F…
 Since i & r combine multiplicatively, they can be added by taking log of the
image intensity,
 so that they can be separated in the FD.
 i variations can be thought as a multiplicative noise & can be reduced by
filtering in the log domain.
 To make the i of an image more even, the HF components are increased and the
LF Components are filtered
 Because the HF are assumed as reflectance in the scene whereas the LF as the
illumination in the scene.i.e.,
 High pass filter is used to suppress LF’s & amplify HF’s in the log intensity
domain.
 i component tends to vary slowly across the image & the reflectance tends to
vary rapidly.
 Therefore, by applying a FD filter the intensity variation across the image can be
reduced while highlighting detail.

15
 H.F….
• Z(x,y) = ln[f(x,y)] = ln[i(x,y)] + ln[r(x,y)] eq-1
• DFT[z(x,y)]
• = DFT{ln[f(x,y)]}
• = DFT{ln[i(x,y)] + ln[r(x,y)]}
• = DFT{ln[i(x,y)]} + DFT{ln[r(x,y)]} eq-2

• Since DFT[f(x,y)] = F(u,v), eq-2 becomes,

• Z(u,v) = Fi(u,v) + Fr(u,v) eq-3

• The function Z represents the FT of the sum of two images: a low

frequency illumination image & a high frequency reflectance image

16
 H.F….
• Thus ,FT of o/p by multiplying the DFT of the i/p with the filter H(u,v).
i.e., S(u,v) = H(u,v) Z(u,v) eq-4
• where S(u,v) is the FT of o/p. Substitute eq-3 in 4,
• we get S(u,v) = H(u,v) [ Fi(u,v) + Fr(u,v) ]
• = H(u,v) Fi(u,v) + H(u,v) Fr(u,v) eq-5
• Applying IDFT to eq-6,
• we get, T -1 [S(u,v)] = T-1 [ H(u,v) Fi(u,v) + H(u,v) Fr(u,v)]
• = T-1 [ H(u,v) Fi(u,v)] + T-1 [H(u,v) Fr(u,v)]
• s(x,y) = i’(x,y) + r’(x,y) eq-6
• The Enhanced image is obtained by taking exponential of the IDFT
s(x,y), y) = e s ( x , y ) = ei '( x , y ) e r '( x , y ) =i ( x, y)r ( x,
g( x,i.e.,
y)
io(x,y) = e i’(x,y) , ro(x,y) = e r’(x,y) 0 0

Where, are the I & r components of the enhanced o/p

17
 Homomorphic Filtering

g( x, y) = e s ( x , y ) = ei '( x , y ) e r '( x , y ) =i ( x, y)r ( x,

y)
0 0

18
 Homomorphic
Filtering
● Illumination Component
– Slow Spatial Variations
– Low Frequencies  log of illumination
– attenuate contributions by illumination

●ReflecLta∈nc1e
– Varies abruptly – junctions of dissimilar objects
– High frequencies  log of reflectance
Component
– amplify contributions by reflectance
● Simultaneous dynamic range compression & contrast
enhanHce>m1ent
19
Homomorphic Filtering

[
( -  ) 1-- c[e2D 02
H (u,v) L
]
 /D ]
+(u,v)
L
= H

20
Image: 1 62x746
γL=0.25, γH=2, c=1, D0=80
21
Band-reject & Band-pass
Filters

H BP (u, v) =1 - H BR (u,
22
 Band-reject &
Band-pass Filters

23
Notch Filters – Narrow Filtering

H NR (u, v) = H k (u, v)H - k (u, v)

k
=1

[
Dk (u, v) (u -M / 2 - uk ) + (v - N / 2 - v )
2
k ]
2 1/
2
=
[
D- (u, v) (u M / 2 + u) 2
+ (v
N / 2 + v)2 1/
]
k k 2
=k - -
24
 Butterworth Notch Reject Filters

H NR (u, v) =
3
 1  1 
 
k =1 1 + [ / D (u, 2n 
 1 + [ / D (u, 2n 
0k k 
 0k - 
 D v)]  D v) ]k

H NP (u, v) =1 - H NR (u,
v) 25
D0=80, n=4

26
27
28
Thank you

29