ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters...
-
Upload
homer-williams -
Category
Documents
-
view
227 -
download
3
Transcript of ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters...
![Page 1: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/1.jpg)
ECE 472/572 - Digital Image Processing
Lecture 5 - Image Enhancement - Frequency Domain Filters09/13/11
![Page 2: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/2.jpg)
2
Roadmap
Introduction– Image format (vector vs. bitmap)– IP vs. CV vs. CG– HLIP vs. LLIP– Image acquisition
Perception– Structure of human eye
• rods vs. conss (Scotopic vision vs. photopic vision)
• Fovea and blind spot• Flexible lens (near-sighted vs. far-
sighted)
– Brightness adaptation and Discrimination
• Weber ratio• Dynamic range
– Image resolution• Sampling vs. quantization
Image enhancement– Enhancement vs. restoration– Spatial domain methods
• Point-based methods– Log trans. vs. Power-law
• Gamma correction• Dynamic range compression
– Contrast stretching vs. HE• What is HE?• Derivation of tran. func.
– Gray-level vs. Bit plane slicing– Image averaging (principle)
• Mask-based (neighborhood-based) methods - spatial filter
– Smoothing vs. Sharpening filter– Linear vs. Non-linear filter– Smoothing
• Average vs. weighted average• Average vs. Median
– Sharpening• UM vs. High boosting• 1st vs. 2nd derivatives
– Frequency domain methods
![Page 3: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/3.jpg)
3
Questions
In-depth understanding– Why do we need to conduct image processing in the frequency
domain?– What does Fourier series do?– What does the Fourier spectrum of an image tell you?– How to calculate the fundamental frequency?– Why is padding necessary?
Properties– Is FT a linear or nonlinear process?– What would the FT of a rotated image look like?– When implementing FFT, what kind of properties are used?– What does the autocorrelation of an image tell you?– What is F(0,0)? Or Why is the center of the FT extremely bright?
![Page 4: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/4.jpg)
4
Why FT? – 1
![Page 5: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/5.jpg)
5
Why FT? – 2
![Page 6: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/6.jpg)
6
Another example
![Page 7: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/7.jpg)
7
Fourier series
Fourier series (SF) can represent any function over a finite interval TF
Outside TF, SF repeats periodically with period TF.
![Page 8: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/8.jpg)
8
Fourier series (cont’)
TF is the interval of signal s(t) over which the Fourier series represents
fF = 1/TF is the fundamental frequency of the Fourier series representation
n is called the “harmonic number”– E.g., 2fF is the second harmonic of the fundamental frequency fF.
The Fourier series representation is always periodic and is linear combinations of sinusoids at fF and its harmonics.
![Page 9: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/9.jpg)
9
Fourier transform
Describe the frequency distribution
nfF
cn
u
v
(0,0)
![Page 10: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/10.jpg)
10
![Page 11: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/11.jpg)
11
1-D Fourier transform
Fourier transform:
Inverse FT:
Complex form
Fourier spectrum Power spectrum
(spectral density) Phase angle
DFT
![Page 12: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/12.jpg)
12
2-D Fourier transform
CFT
DFT
![Page 13: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/13.jpg)
13
Understanding and implementing Fourier transform
(0,0)
f(x,y)
x
y(0,0)
|F(u,v)|
u
v
![Page 14: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/14.jpg)
14
![Page 15: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/15.jpg)
15
Understanding and implementing Fourier transform
According to “translation”
(0,0)
f(x,y)
x
y
-255
255
(0,0)
f(x,y)(-1)x+y
x
y
-0
0
(0,0)
|F(u-M/2,v-N/2)|u
v
![Page 16: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/16.jpg)
16
Understanding and implementing Fourier transform
![Page 17: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/17.jpg)
17
Block diagram of FTIFT
f(x,y)(-1)x+yg(x,y)(-1)x+y
![Page 18: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/18.jpg)
18
Impulse transforms
![Page 19: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/19.jpg)
19
Typical transforms
Gaussian hump Gaussian hump
Rectangular (square aperture) sinc
Pillbox (circular aperture) jinc
![Page 20: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/20.jpg)
20
Typical transforms
Gaussian ridge
Line impulse
![Page 21: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/21.jpg)
21
Reference
All figures scanned from R. N. Bracewell’s “Two-Dimensional Imaging,” Prentice Hall, 1995.
![Page 22: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/22.jpg)
22
2D FT pairs
![Page 23: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/23.jpg)
23
Important properties of FT
Linearity (distributivity && scaling) Separability Translation Periodicity Conjugate symmetry Rotation Convolution Correlation Sampling
![Page 24: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/24.jpg)
24
Linearity
FT is a linear image processing method
Linear System
x1(t) y1(t)
x2(t) y2(t)
a*x1(t) + b*x2(t) a*y1(t) + b*y2(t)
![Page 25: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/25.jpg)
25
Separability
f(x, y) F(x, v)
Rowtransform F(u, v)
Columntransform
![Page 26: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/26.jpg)
26
Translation
![Page 27: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/27.jpg)
27
Periodicity and Conjugate symmetry
Periodicity
Conjugate symmetry
![Page 28: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/28.jpg)
28
Example
![Page 29: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/29.jpg)
29
Rotation
![Page 30: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/30.jpg)
30
Averaging
![Page 31: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/31.jpg)
31
Convolution
Continuous and discrete convolution
The convolution theorem
Practically, computing the discrete convolution in the frequency domain often is more efficient than doing it in the spatial domain directly
![Page 32: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/32.jpg)
32
Correlation
Continuous and discrete correlation
The correlation theorem
![Page 33: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/33.jpg)
33
Correlation (cont’)
Autocorrelation vs. cross correlation Autocorrelation theorem
Application: template or prototype matching
![Page 34: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/34.jpg)
34
Practical issues – Implement convolution in frequency domain
In spatial domain
In frequency domain– f*g F(f)G(g)– Phase? Mag?– How to pad?
![Page 35: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/35.jpg)
35
![Page 36: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/36.jpg)
36
![Page 37: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/37.jpg)
37
![Page 38: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/38.jpg)
38
![Page 39: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/39.jpg)
39
Difference image from convolution in the spatial domain
Convolution in the frequency domain
Conv. spatially
No padding With padding
![Page 40: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/40.jpg)
40
Different enhancement approaches
Lowpass filterHighpass filterHomomorphic filter
![Page 41: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/41.jpg)
41
Lowpass filtering
Ideal filter– D(u, v): distance from point (u, v) to the origin– cutoff frequency (D0)– nonphysical– radially symmetric about the origin
Butterworth filter
Gaussian lowpass filter
![Page 42: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/42.jpg)
42
![Page 43: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/43.jpg)
43
Power ratio99.9699.6599.0497.84
![Page 44: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/44.jpg)
44
Highpass filter
Ideal filter
Butterworth filter
Gaussian highpass filter
![Page 45: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/45.jpg)
45
![Page 46: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/46.jpg)
46
![Page 47: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/47.jpg)
47
Example
![Page 48: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/48.jpg)
48
The Laplacian in the frequency domain
Pay attention to the scalingfactor
![Page 49: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/49.jpg)
49
UM in the frequency domain
![Page 50: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/50.jpg)
50
Homomorphic filtering
A simple image model– f(x,y): the intensity is called the gray level for
monochrome image– f(x, y) = i(x, y).r(x, y)– 0 < i(x, y) < inf, the illumination– 0< r(x, y) < 1, the reflectance
![Page 51: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/51.jpg)
51
Homomorphic filter (cont’)
![Page 52: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/52.jpg)
52
Homomorphic filter (cont’)
The illumination component– Slow spatial variations– Low frequency
The reflectance component– Vary abruptly, particularly at the junctions of
dissimilar objects– High frequency
Homomorphic filters– Affect low and high frequencies differently– Compress the low frequency dynamic range– Enhance the contrast in high frequency
![Page 53: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/53.jpg)
53
Homomorphic Filter (cont’)
![Page 54: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/54.jpg)
54
![Page 55: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/55.jpg)
55
Homomorphic filter - example
![Page 56: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/56.jpg)
56
Point processing
Simple gray level transformations– Image negatives
– Log transformations
– Power-law transformations
– Contrast stretching
– Gray-level slicing
– Bit-plane slicing
Histogram processing– Histogram
equalization
– *Histogram matching (specification)
Arithmetic/logic operations– Image averaging
Mask processing (spatial filters)
Smoothing filters (blur details)– Average, weighted
average
– Order statistics (e.g. median)
Sharpening filters (highlight details)– Unsharp masking
– High-boost filters
– Derivative filters• The Laplacian
• The Gradient
• Frequency domain filters
• Smoothing filters (blur details)
• Ideal lowpass filter• Butterworth lowpass• Gaussian lowpass
• Sharpening filters (highlight details)
– Unsharp masking– High-boost filters– Derivative filters - The
Laplacian
– Ideal highpass filter– Butterworth highpass filter– Gaussian highpass filter
• Homomorphic filtering
![Page 57: ECE 472/572 - Digital Image Processing Lecture 5 - Image Enhancement - Frequency Domain Filters 09/13/11.](https://reader035.fdocuments.in/reader035/viewer/2022062320/56649d1f5503460f949f3ec5/html5/thumbnails/57.jpg)
57
FFT and IFFT