sparse image representation via combined transforms - Convex ...
sparse image representation via combined transforms - Convex ... sparse image representation via combined transforms - Convex ...
32 CHAPTER 3. IMAGE TRANSFORMS AND IMAGE FEATURES the set of eigenvalues of this system. Note: when the signal is infinitely long and the impulse response is relatively short, we can drop the cyclic constraint. Since an LTI system is a widely assumed model in signal processing, the Fourier transform plays an important role in analyzing this type of system. 2. A covariance matrix of a stationary Gaussian (i.e., Normal) time series is Toeplitz and symmetric. If we can assume that the time series does not have long-range dependency 2 , which implies that off-diagonal elements diminish when they are far from the diagonal, then the covariance matrix is almost circulant. Hence, the Fourier series again become the eigenvectors of this matrix. The DFT matrix DFT N nearly diagonalizes the covariance matrix, which implies that after the discrete Fourier transform of the time series, the coefficients are almost independently distributed. This idea can be depicted as DFT N ·{the covariance matrix}·DFT T N ≈{a diagonal matrix}. Generally speaking, it is easier to process independent coefficients than a correlated time series. The eigenvalue amplitudes determine the variance of the associated coefficients. Usually, the eigenvalue amplitudes decay fast: say, exponentially. The coefficients corresponding to the eigenvalues having large amplitudes are important coefficients in analysis. The remaining coefficients could almost be considered constant. Some research in this direction can be found in [58], [59] and [101]. From the above two examples, we see that FT plays an important role in both DSP and time series analysis. Fast Fourier Transform To illustrate the importance of the fast Fourier transform in contemporary computational science, we find that the following sentences (the first paragraph of the Preface in [136]) say exactly what we would like to say. The fast Fourier transform (FFT) is one of the truly great computational developments of this century. It has changed the face of science and engineering so 2 No long-range dependency means that if two locations are far away in the series, then the corresponding two random variables are nearly independent.
3.1. DCT AND HOMOGENEOUS COMPONENTS 33 much so that it is not an exaggeration to say that life as we know it would be very different without the FFT. — Charles Van Loan [136]. There are many ways to derive the FFT. The following theorem, which is sometimes called Butterfly Theorem, seems the most understandable way to describe why FFT works. The author would like to thank Charles Chui [30] for being the first fellow to introduce this theorem to me. The original idea started from Cooley and Tukey [32]. To state the theorem, let us first give some notation for matrices. Let P 1 and P 2 denote two permutation matrices satisfying for any mn-dimensional vector x =(x 0 ,x 1 ,... ,x mn−1 ) ∈ R mn ,wehave, ⎛ P 1 ⎜ ⎝ x 0 x 1 . x mn−1 ⎛ ⎞ ⎟ = ⎠ mn×1 ⎜ ⎝ ⎛ ⎜ ⎝ ⎛ ⎜ ⎝ ⎛ ⎜ ⎝ x 0 x n . x (m−1)n x 1 x 1+n . x 1+(m−1)n x (n−1) ⎞ ⎟ ⎠ • • • x (n−1)+n . ⎞ ⎟ ⎠ m×1 m×1 ⎞ ⎟ ⎠ ⎞ ; (3.3) ⎟ ⎠ x (n−1)+(m−1)n m×1
- Page 9: Abstract We consider sparse image d
- Page 12 and 13: xii
- Page 14 and 15: 2.3 Discussion.....................
- Page 16 and 17: 6 Simulations 119 6.1 Dictionary...
- Page 18 and 19: xviii
- Page 21 and 22: List of Figures 2.1 Shannon’s sch
- Page 23 and 24: A.3 Edgelet transform of the wood g
- Page 25 and 26: Nomenclature Special sets N .......
- Page 27 and 28: List of Abbreviations BCR .........
- Page 29 and 30: Chapter 1 Introduction 1.1 Overview
- Page 31 and 32: Chapter 2 Sparsity in Image Coding
- Page 33 and 34: 2.1. IMAGE CODING 5 INFORMATION SOU
- Page 35 and 36: 2.1. IMAGE CODING 7 2.1.2 Source an
- Page 37 and 38: 2.1. IMAGE CODING 9 x ✲ T y ERROR
- Page 39 and 40: 2.1. IMAGE CODING 11 where Q stands
- Page 41 and 42: 2.2. SPARSITY AND COMPRESSION 13 Pr
- Page 43 and 44: 2.2. SPARSITY AND COMPRESSION 15 av
- Page 45 and 46: 2.2. SPARSITY AND COMPRESSION 17 wi
- Page 47 and 48: 2.2. SPARSITY AND COMPRESSION 19 lo
- Page 49 and 50: 2.3. DISCUSSION 21 tail compact. Th
- Page 51 and 52: 2.4. PROOF 23 The index l does not
- Page 53 and 54: Chapter 3 Image Transforms and Imag
- Page 55 and 56: 27 Some of the figures show the bas
- Page 57 and 58: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 59: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 63 and 64: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 65 and 66: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 67 and 68: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 69 and 70: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 71 and 72: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 73 and 74: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 75 and 76: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 77 and 78: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 79 and 80: 3.1. DCT AND HOMOGENEOUS COMPONENTS
- Page 81 and 82: 3.2. WAVELETS AND POINT SINGULARITI
- Page 83 and 84: 3.2. WAVELETS AND POINT SINGULARITI
- Page 85 and 86: 3.2. WAVELETS AND POINT SINGULARITI
- Page 87 and 88: 3.2. WAVELETS AND POINT SINGULARITI
- Page 89 and 90: 3.2. WAVELETS AND POINT SINGULARITI
- Page 91 and 92: 3.2. WAVELETS AND POINT SINGULARITI
- Page 93 and 94: 3.3. EDGELETS AND LINEAR SINGULARIT
- Page 95 and 96: 3.4. OTHER TRANSFORMS 67 uncertaint
- Page 97 and 98: 3.4. OTHER TRANSFORMS 69 Chirplets
- Page 99 and 100: 3.4. OTHER TRANSFORMS 71 Folding. A
- Page 101 and 102: 3.4. OTHER TRANSFORMS 73 We can app
- Page 103 and 104: 3.5. DISCUSSION 75 give only a few
- Page 105 and 106: 3.7. PROOFS 77 the ijth component o
- Page 107 and 108: 3.7. PROOFS 79 Similarly, we have [
- Page 109 and 110: Chapter 4 Combined Image Representa
3.1. DCT AND HOMOGENEOUS COMPONENTS 33<br />
much so that it is not an exaggeration to say that life as we know it would be<br />
very different without the FFT. — Charles Van Loan [136].<br />
There are many ways to derive the FFT. The following theorem, which is sometimes<br />
called Butterfly Theorem, seems the most understandable way to describe why FFT works.<br />
The author would like to thank Charles Chui [30] for being the first fellow to introduce this<br />
theorem to me. The original idea started from Cooley and Tukey [32].<br />
To state the theorem, let us first give some notation for matrices. Let P 1 and P 2 denote<br />
two permutation matrices satisfying for any mn-dimensional vector x =(x 0 ,x 1 ,... ,x mn−1 )<br />
∈ R mn ,wehave,<br />
⎛<br />
P 1 ⎜<br />
⎝<br />
x 0<br />
x 1<br />
.<br />
x mn−1<br />
⎛<br />
⎞<br />
⎟ =<br />
⎠<br />
mn×1<br />
⎜<br />
⎝<br />
⎛<br />
⎜<br />
⎝<br />
⎛<br />
⎜<br />
⎝<br />
⎛<br />
⎜<br />
⎝<br />
x 0<br />
x n<br />
.<br />
x (m−1)n<br />
x 1<br />
x 1+n<br />
.<br />
x 1+(m−1)n<br />
x (n−1)<br />
⎞<br />
⎟<br />
⎠<br />
•<br />
•<br />
•<br />
x (n−1)+n<br />
.<br />
⎞<br />
⎟<br />
⎠<br />
m×1<br />
m×1<br />
⎞<br />
⎟<br />
⎠<br />
⎞<br />
; (3.3)<br />
⎟<br />
⎠<br />
x (n−1)+(m−1)n<br />
m×1