DW Home Page
|Discovering Wavelets, the book
maintained by Edward Aboufadel and Steven Schlicker
You can order the book from an online bookstore.
Here is a review of the book from Zentralblatt Math. There are a few reviews at Amazon.com. Here's a not-so-kind review from Ian Kaplan.
Here is the preface from the book, along with the Table of Contents.
For the past several years, reflecting the excitement and creativity surrounding the subject of wavelets, articles about wavelets have appeared in professional publications, on the World Wide Web, and in mainstream magazines and newspapers. Much of this enthusiasm for wavelets comes from known and from potential applications. For example, wavelets have found use in image processing, in the restoration of recordings, and in seismology.
Many books are available on wavelets, but most are written at such a level that only research mathematicians can read them. The purpose of this book is to make wavelets accessible to anyone with a background in basic linear algebra (for example, graduate and undergraduate students), and to serve as an introduction for the nonspecialist. The level of the applications and the format of this book make it an excellent textbook for an introductory course on wavelets or as a supplement to a first or second course in linear algebra or numerical analysis. Potential readers would be intrigued by the discussion of the Wavelet/Scalar Quantization Standard, currently used by the Federal Bureau of Investigation to compress, transmit, and store images of fingerprints. In Britain, Scotland Yard also uses wavelets for the same purpose.
The projects that are contained within this book allow real applications to be incorporated into the mathematics curriculum. This fits well with the current trend of infusing mathematics courses with applications; an approach which is summarized well by Avner Friedman and John Lavery in their statement about an industrial mathematics program:
"[this approach provides] students an immense opportunity for greater and deeper contributions in all areas of the natural and social sciences, engineering and technology."The practice of either basing a mathematics course on applications or infusing a course with applications has been seen primarily in calculus courses. However, this enthusiasm for real applications has not been as obvious in our upper-level mathematics courses, such as linear algebra, abstract algebra, or number theory, which are taken primarily by mathematics majors. Many of these majors intend to be teachers, and applications such as wavelets can provide breadth to the curriculum. As stated in a recent report of the Mathematical Association of America:
"it is vitally important that [prospective math teachers'] undergraduate experience provide a broad view of the discipline, [since] for the many students who may never make professional use of mathematics, depth through breadth offers a strong base for appreciating the true power and scope of the mathematical sciences."The major benefit of this book is that it presents basic and advanced concepts of wavelets in a way that is accessible to anyone with only a basic knowledge of linear algebra. The discussion of wavelets begins with an idea of Gil Strang to represent the Haar wavelets as vectors in Rn, and is driven by the desire to introduce readers to how the FBI is compressing fingerprint images. Chapter 1 introduces the basic concepts of wavelet theory in a concrete setting: the Haar wavelets along with the problem of digitizing fingerprints. The rest of the book builds on this material. To fully understand the concepts in this chapter, a reader need only have an understanding of basic linear algebra ideas -- matrix multiplication, adding and multiplying vectors by scalars, linear independence and dependence.
Chapter 2 builds on the ideas presented in chapter 1, developing more of the theory of wavelets along with function spaces. Readers get a better sense of how one might deduce the ideas presented in chapter 1. To fully understand the material in this chapter, a reader needs more sophisticated mathematics, such as inner product spaces and projections. The necessary background material from linear algebra that a reader needs to know to fully understand the discussion of wavelets in this book is contained in appendix A.
Chapter 3 features more advanced topics such as filters, multiresolution analysis, Daubechies wavelets, and further applications. These topics are all introduced by comparison to the material developed with the Haar wavelets in chapters 1 and 2. These topics would be of interest to anyone who desires to read some of the more technical books or papers on wavelets, or to anyone seeking a starting point for research projects. There are some new concepts introduced in this chapter (e.g. density, fixed-point algorithms, and L2 spaces). They are discussed in enough detail to allow the reader to understand these concepts and how they relate to wavelets, but not in so much detail that these ancillary topics distract from the major topic of wavelets. For example, the Fourier transform does not appear until the final section of chapter 3.
Chapter 4 contains projects that could be used in linear algebra courses. Along with these projects, some of the problems introduce advanced topics that could be used as starting points for research by undergraduates. There are also appendices that review linear algebra topics and present Maple commands that are useful for some of the problems.
These notes originated in courses (Linear Algebra II) that the authors taught during summers 1996 and 1997 at Grand Valley State University in Allendale, Michigan. Students in each of these courses had previously completed Linear Algebra I, where they learned to solve systems of linear equations, were introduced to matrix and vector operations, and encountered for the first time the fundamental ideas of spanning sets, linear independence and dependence, bases, and dimension. The topics for the second semester course were grouped around two major themes: linear transformations and orthogonal projections. In the 1996 course, about one of every five class meetings was set aside to learn about wavelets, and the material in these notes was timed to coincide with the flow of topics in the rest of the course. (For example, using orthogonal projections to approximate functions with wavelets was done as we studied orthogonal projections in inner product spaces.) Along with the information presented in these notes, various articles were distributed to students as required readings.
In the 1997 course, groups of students submitted written reports based on a subset of the problems in chapters 1 and 2. In addition, each group created a gray-scale image in a 16-by-16 grid of pixels using a program, Pixel Images, written by Schlicker. Each group processed, compressed, and decompressed their image using Haar wavelets and entropy coding. A modified version of this activity is included in chapter 4.
While we initially introduced wavelets into the second semester linear algebra course, many of the tools and concepts can fit equally well in a first semester linear algebra course. In the last two years we have used various approaches in both semesters of linear algebra to expose our students to this valuable and exciting area of mathematics. In all of these courses, the students have expressed that they appreciated seeing a connection between what they were learning in college and what they saw happening in the world.
A note on how to use this book. To learn mathematics it is important to become conversant with the terminology and to actually work some problems. Throughout this book, key terms and phrases are highlighted in italics. For pedagogical purposes, we have included some terminology that is not standard in the literature: the phrase daughter wavelets in chapter 1 is used to describe the functions that are obtained from dilations and translations of the mother wavelet, son wavelets in chapter 2 describes the functions that are obtained from dilations and translations of the father wavelet or scaling function, and image box in chapter 2 refers to any figure that contains projections and residuals of an original image after processing with wavelets. We feel these terms provide appropriately descriptive labels and use them without hesitation.
The problems posed in this text are distributed throughout the reading rather than at the end of each chapter. This is important because completing the problems is vital to learning about wavelets. In fact, it is necessary to solve those problems marked with bold numbers to completely understand the text. Also, a computer algebra system is necessary to complete some problems, and hints and an appendix are provided for those readers who have access to Maple. Answers to selected problems are provided in an appendix. Pixel Images and Maple worksheets are available at our web site:
Table of Contents
Chapter 1 Wavelets, Fingerprints, and Image Processing
section 1.1 Problems of the Digital Age
section 1.2 Digitizing Fingerprint Images
section 1.3 Signals
section 1.4 The Haar Wavelet Family
section 1.5 Processing Signals
section 1.6 Thresholding and Compression of Data
section 1.7 The FBI Wavelet/Scalar Quantization Standard
Chapter 2 Wavelets and Orthogonal Decompositions
section 2.1 A Lego World
section 2.2 The Wavelet Sons
section 2.3 Sibling Rivalry: Two Bases for Vn
section 2.4 Averaging and Differencing
section 2.5 Projecting Functions Onto Wavelet Spaces
section 2.6 Function Processing and Image Boxes
section 2.7 A Summary of Two Approaches to Wavelets
Chapter 3 Multiresolutions, Cascades, and Filters
section 3.1 Extending the Haar Wavelets to the Real Line
section 3.2 Other Elementary Wavelet Families
section 3.3 Multiresolution Analysis
section 3.4 The Haar Scaling Function Rediscovered
section 3.5 Relationships Between the Mother and Father Wavelets
section 3.6 Daubechies Wavelets
section 3.7 High and Low Pass Filters
section 3.8 More Problems of the Digital Age: Compact Discs
Chapter 4 Sample Projects
section 4.1 Introduction: Overview of Projects
section 4.2 Linear Algebra Project: Image Processing and Compression
section 4.3 A Wavelet-Based Search Engine
section 4.4 B-Splines
section 4.5 Processing with the D4 Wavelets
section 4.6 Daubechies Wavelets with Six Refinement Coefficients
Appendix A Vector Spaces and Inner Product Spaces
section A.1 Vector Spaces
section A.2 Subspaces
section A.3 Inner Product Spaces
section A.4 The Orthogonal Decomposition Theorem
Appendix B Maple Routines
section B.1 Matrix Generator
section B.2 Processing Sampled Data
section B.3 Projections onto Wavelet Spaces
section B.4 The Cascade Algorithm
section B.5 Processing an Image from Pixel Images
Appendix C Answers to Selected Problems
Appendix D Glossary of Symbols