recap of monday linear filtering convolution differential filters filter types boundary conditions
TRANSCRIPT
![Page 1: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/1.jpg)
Recap of Monday
linear Filteringconvolutiondifferential filtersfilter typesboundary conditions.
![Page 2: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/2.jpg)
The Frequency Domain (Szeliski 3.4)
CS129: Computational PhotographyJames Hays, Brown, Spring 2011
Somewhere in Cinque Terre, May 2005
Slides from Steve Seitzand Alexei Efros
![Page 3: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/3.jpg)
Salvador Dali“Gala Contemplating the Mediterranean Sea, which at 30 meters becomes the portrait of Abraham Lincoln”, 1976
![Page 4: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/4.jpg)
![Page 5: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/5.jpg)
![Page 6: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/6.jpg)
A nice set of basis
This change of basis has a special name…
Teases away fast vs. slow changes in the image.
![Page 7: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/7.jpg)
Jean Baptiste Joseph Fourier (1768-1830)
had crazy idea (1807):Any periodic function can be rewritten as a weighted sum of sines and cosines of different frequencies.
Don’t believe it? • Neither did Lagrange,
Laplace, Poisson and other big wigs
• Not translated into English until 1878!
But it’s true!• called Fourier Series
![Page 8: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/8.jpg)
A sum of sinesOur building block:
Add enough of them to get any signal f(x) you want!
How many degrees of freedom?
What does each control?
Which one encodes the coarse vs. fine structure of the signal?
xAsin(
![Page 9: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/9.jpg)
Fourier TransformWe want to understand the frequency of our signal. So, let’s reparametrize the signal by instead of x:
xAsin(
f(x) F()Fourier Transform
F() f(x)Inverse Fourier Transform
For every from 0 to inf, F() holds the amplitude A and phase of the corresponding sine
• How can F hold both? Complex number trick!
)()()( iIRF 22 )()( IRA
)(
)(tan 1
R
I
We can always go back:
![Page 10: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/10.jpg)
Time and Frequency
example : g(t) = sin(2pf t) + (1/3)sin(2p(3f) t)
![Page 11: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/11.jpg)
Time and Frequency
example : g(t) = sin(2pf t) + (1/3)sin(2p(3f) t)
= +
![Page 12: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/12.jpg)
Frequency Spectra
example : g(t) = sin(2pf t) + (1/3)sin(2p(3f) t)
= +
![Page 13: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/13.jpg)
Frequency SpectraUsually, frequency is more interesting than the phase
![Page 14: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/14.jpg)
= +
=
Frequency Spectra
![Page 15: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/15.jpg)
= +
=
Frequency Spectra
![Page 16: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/16.jpg)
= +
=
Frequency Spectra
![Page 17: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/17.jpg)
= +
=
Frequency Spectra
![Page 18: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/18.jpg)
= +
=
Frequency Spectra
![Page 19: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/19.jpg)
= 1
1sin(2 )
k
A ktk
Frequency Spectra
![Page 20: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/20.jpg)
Frequency Spectra
![Page 21: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/21.jpg)
Extension to 2D
in Matlab, check out: imagesc(log(abs(fftshift(fft2(im)))));
![Page 22: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/22.jpg)
Man-made Scene
![Page 23: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/23.jpg)
Can change spectrum, then reconstruct
![Page 24: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/24.jpg)
Low and High Pass filtering
![Page 25: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/25.jpg)
The Convolution Theorem
• The Fourier transform of the convolution of two functions is the product of their Fourier transforms
• Convolution in spatial domain is equivalent to multiplication in frequency domain!
]F[]F[]F[ hghg
![Page 26: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/26.jpg)
2D convolution theorem example
*
f(x,y)
h(x,y)
g(x,y)
|F(sx,sy)|
|H(sx,sy)|
|G(sx,sy)|
![Page 27: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/27.jpg)
Fourier Transform pairs
![Page 28: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/28.jpg)
Low-pass, Band-pass, High-pass filters
low-pass:
High-pass / band-pass:
![Page 29: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/29.jpg)
Edges in images
![Page 30: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/30.jpg)
What does blurring take away?
original
![Page 31: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/31.jpg)
What does blurring take away?
smoothed (5x5 Gaussian)
![Page 32: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/32.jpg)
High-Pass filter
smoothed – original
![Page 33: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/33.jpg)
Band-pass filtering
Laplacian Pyramid (subband images)Created from Gaussian pyramid by subtraction
Gaussian Pyramid (low-pass images)
![Page 34: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/34.jpg)
Laplacian Pyramid
How can we reconstruct (collapse) this pyramid into the original image?
Need this!
Originalimage
![Page 35: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/35.jpg)
Why Laplacian?
Laplacian of Gaussian
Gaussian
delta function
![Page 36: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/36.jpg)
Image gradientThe gradient of an image:
The gradient points in the direction of most rapid change in intensity
The gradient direction is given by:
• how does this relate to the direction of the edge?
The edge strength is given by the gradient magnitude
![Page 37: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/37.jpg)
Effects of noise
Consider a single row or column of the image• Plotting intensity as a function of position gives a signal
Where is the edge?
How to compute a derivative?
![Page 38: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/38.jpg)
Where is the edge?
Solution: smooth first
Look for peaks in
![Page 39: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/39.jpg)
Derivative theorem of convolution
This saves us one operation:
![Page 40: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/40.jpg)
Laplacian of Gaussian
Consider
Laplacian of Gaussianoperator
Where is the edge? Zero-crossings of bottom graph
![Page 41: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/41.jpg)
2D edge detection filters
is the Laplacian operator:
Laplacian of Gaussian
Gaussian derivative of Gaussian
![Page 42: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/42.jpg)
Campbell-Robson contrast sensitivity curveCampbell-Robson contrast sensitivity curve
![Page 43: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/43.jpg)
Depends on Color
R G B
![Page 44: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/44.jpg)
Lossy Image Compression (JPEG)
Block-based Discrete Cosine Transform (DCT)
![Page 45: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/45.jpg)
Using DCT in JPEG
The first coefficient B(0,0) is the DC component, the average intensity
The top-left coeffs represent low frequencies, the bottom right – high frequencies
![Page 46: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/46.jpg)
Image compression using DCT
DCT enables image compression by concentrating most image information in the low frequencies
Lose unimportant image info (high frequencies) by cutting B(u,v) at bottom right
The decoder computes the inverse DCT – IDCT
•Quantization Table3 5 7 9 11 13 15 17
5 7 9 11 13 15 17 19
7 9 11 13 15 17 19 21
9 11 13 15 17 19 21 23
11 13 15 17 19 21 23 25
13 15 17 19 21 23 25 27
15 17 19 21 23 25 27 29
17 19 21 23 25 27 29 31
![Page 47: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/47.jpg)
JPEG compression comparison
89k 12k
![Page 48: Recap of Monday linear Filtering convolution differential filters filter types boundary conditions](https://reader035.vdocuments.us/reader035/viewer/2022062314/56649f535503460f94c77f31/html5/thumbnails/48.jpg)
Summary
Frequency domain can be useful for
Analysis
Computational efficiency
Compression