Statistical Image Processing and Multidimensional Modeling

List of Errata |

Chapter | Page | Location in Text | Nature of Error | |
---|---|---|---|---|

2 | 14 | Table 2.2 | Phrase "those are" accidentally appears twice in the table caption. | |

2 | 27 | Section 2.3 | In (2.28), the induced norm on matrix C is based on the norm of Cx, not Ax. | |

2 | 34 | Section 2.4.1 | In the second sentence, it is *uniqueness* that would be satisfied, not existence. | |

2 | 53 | Problem 2.2 | Part (c) should ask you to prove that *existence* fails, not uniqueness. | |

3 | 73 | Section 3.2.4 | The added mean should appear in the equation preceding (3.83), and not in (3.84). The three equations should therefore look like this. | |

4 | 89 | Section 4.1.1 | In (4.13), P is the covariance of _{o}z(0), which is a random vector, not of z, which is just a constant vector.
_{o} | |

5 | 146 | Section 5.3.3 | In (5.23), the number of bands in the three braces should be 1, b, b, rather than 1, ^{2}b, ^{2}b.
^{4} | |

5 | 157 | Section 5.5.2 | In Fig. 5.13, the very first row of the array is slightly misleading. The (2 -1) constraint is a second-order constraint with a "missing" boundary value of zero; however really this does not constrain the curve to tend to zero at the boundary, rather (2 -1) is a slope constraint at the boundary. | |

8 | 267 | Algorithm 3 | Not quite an "error", but definitely misleading. The `real` command is needed only after the `ifft` projecting back to the spatial domain. We would normally not take the real part after the `fft` projecting into the frequency domain. I do this twice (middle and bottom) in Algorithm 3.
| |

9 | 304 | Figure 9.4 | The right panel corresponds to w > 1, not to w < 1 | |

9 | 319 | Algorithm 9 | In the line calculating "bc", rather than "f*" it should be function "subfn" that is called. | |

9 | 324 | Problem 9.3(c) | should actually be |

(Page last updated December 6, 2016, [an error occurred while processing this directive])