Download presentation
Presentation is loading. Please wait.
Published byPaul Rose Modified over 9 years ago
1
7- 1 Chapter 7: Fourier Analysis Fourier analysis = Series + Transform ◎ Fourier Series -- A periodic (T) function f(x) can be written as the sum of sines and cosines of varying amplitudes and frequencies
2
7- 2 ○ Some function is formed by a finite number of sinuous functions
3
7- 3 Some function requires an infinite number of sinuous functions to compose
4
7- 4 Spectrum The spectrum of a periodic function is discrete, consisting of components at dc, 1/T, and its multiples, e.g., For non-periodic functions, i.e., The spectrum of the function is continuous
5
7- 5 ○ In complex form: Compare with
6
7- 6 Euler ’ s formula:
7
7- 7
8
7- 8
9
7- 9 Continuous case
10
7- 10 Discrete case: ◎ Fourier Transform
11
7- 11 Matrix form Let
12
7- 12 。 Example: f = {1,2,3,4}. Then, N = 4,
13
7- 13
14
7- 14 ○ Inverse DFT Let
15
7- 15 。 Example:
16
7- 16 ◎ Properties ○ Linearity: e.g., Noise removal f’ = f + n, n: additive noise,
17
8-17 Fourier spectrum noise Corresponding spatial noise
18
○ Scaling : Show:
19
7- 19 ○ Periodicity:
20
7- 20 ○ Shifting:
21
7- 21 。 Example:
22
7- 22 ◎ Convolution: Convolution theorem: Correlation theorem ◎ Correlation
23
7- 23 。 Discrete Case: A = 4, B = 5, A + B – 1 = 8, e.g.,
24
7- 24 * Convolution can be defined in terms of polynomial product Extend f, g to if f, g have different numbers of sample points Let Compute The coefficients of to form the result of convolution
25
7- 25 。 Example: Let The coefficients of form the convolution
26
7- 26
27
7- 27 ○ Fast Fourier Transform (FFT) -- Successive doubling method
28
7- 28
29
7- 29 。 Time complexity : the length of the input sequence FT: FFT: Times of speed increasing: N FT FFT Ratio 4 16 8 2.0 8 84 24 2.67 16 256 64 4.0 32 1024 160 6.4 64 4096 384 10.67 128 16384 896 18.3 256 65536 2048 32.0 512 262144 4608 56.9 1024 1048576 10240 102.4
30
7- 30 ○ Inverse FFT ← Given ← compute i. Input into FFT. The output is ii. Taking the complex conjugate and multiplying by N, yields the f(x)
31
7- 31 ◎ 2D Fourier Transform ○ FT: IFT:
32
7- 32 ◎ Properties ○ Filtering: every F(u,v) is obtained by multiplying every f(x,y) by a fixed value and adding up the results. DFT can be considered as a linear filtering ○ DC coefficient:
33
7- 33 ○ Separability:
34
7- 34 ○ Conjugate Symmetry: F(u,v) = F*(-u,-v)
35
7- 35 ○ Shifting
36
7- 36 ○ Rotation Polor coordinates:
37
7- 37 ○ Display: effect of log operation
38
7- 38
39
7- 39 ◎ Image Transform
40
7- 40 ◎ Filtering in Frequency Domain ○ Low pass filtering I FT m IFT
41
7- 41 D = 5 D = 30 ○ High pass filtering
42
7- 42 Different Ds
43
7- 43 ◎ Butterworth Filtering ○ Low pass filter ○ High pass filter
44
7- 44 ○ Low pass filter ○ High pass filter
45
7- 45 ◎ Homomorphic Filtering -- Deals with images with large variation of illumination, e.g., sunshine + shadows -- Both reduce intensity range and increases local contrast ○ Idea: I = LR, L: illumination, R: Reflectance logI = logL + logR low frequency high frequency
46
7- 46
47
7- 47
48
7- 48 ○ Fast Fourier Transform (FFT) -- Successive doubling method Assume Let Let N = 2M.
49
7- 49 = ] ∵ = ]
50
7- 50 Let --------- (B) Consider
51
7- 51
52
7- 52 ---- (C)
53
7- 53 F(u+M) = Recursively divide F(u) and F(u+M), ○ Analysis : The Fourier sequence F(u), u = 0, …, N-1 of f(x), x = 0, …, N-1 can be formed from sequences F(u) = Eventually, each contains one element F(w), i.e., w = 0, and F(w) = f(x). u = 0, ……, M-1
54
7- 54
55
7- 55 ○ Example: needs { f(0), f(2), f(4), f(6) } needs { f(1), f(3), f(5), f(7) } Computing Input { f(0), f(1), ……, f(7) }
56
7- 56 Reorder the input sequence into {f(0), f(4), f(2), f(6), f(1), f(5), f(3), f(7)} Computing * Bit-Reversal Rule
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.