errata
Typos in Discrete Fourier Analysis and Wavelets: Applications to Signal and Image Processing
The following typos were found during 2 semesters in which the book was used at IME/USP.
- p. 104, Exercise 2.19, line 3 should read "(...) coefficients âk, l, âm − k, l, âk, n − l, and âm − k, n − l.".
- p. 126, Example 3.1, all data from B̂ (B̂, q(B̂), B̃, and error values) are wrong. The corresponding code is on Octave examples from the class of 21/09.
- p. 146, 4th paragraph, 3rd line should read "(...) equation (2.6) in Chapter 2.".
- p. 165, the formula for Y(z) should be expanded as − 8z3 − 4z2 − 2z.
- p. 184, 5th paragraph should read "(...) zeros out all components xk with k > m + M − 1".
- p. 186, last paragraph should read "Equation (5.6) makes it obvious that the N − point DFT Y is not identical to X but is distorted by convolution with W, the Fourier Transform of the window vector.".
- p. 207, 2nd paragraph should read "In general, the kth component of vl is (1 / 2)(xk + xk − 1) if k is even and (1 / 2)(xk + 1 + xk) if k is odd. The kth component of vh is (1 / 2)(xk − xk − 1) if k is even (...)".
- p. 213, 3rd paragraph, 2nd line the equations have double opening parenthesis.
- p. 222, 4th paragraph should read "But within the range 0 ≤ m ≤ N − 1 equations (6.15) only require knowledge of U(Xl) and U(Xh) on at most the range − N / 2 ≤ j ≤ N / 2."
- p. 224, first paragraph of section 6.5.3 should read: "If all convolutions in the finite case are circular, then Xl = DMlax and Xh = DMhax, where Mla and Mha are the circulant matrices for the analysis filters and D is a matrix that represents downsampling.
- p. 270, Figure 7.3, figures seems to be mirrored. They are synthesized like this if we switch filters of analysis and synthesis. Compare the situation of figures 7.2 e 7.4 (mirrored) with figures 7.13 e 7.14.
- p. 274, Equation 7.6, there should be no 2k / 2 there.
- p. 288, last line of Equation 7.30, missing + δ0 on the sum.