Computer vision. Algorithms and applications (Q5893947)

The aim of this book is to provide a course in computer vision for undergraduate students in computer science or electrical engineering. The author has given several such courses in a project-oriented way. The focus is on algorithms and applications. In the author's own words: ``Thus, this book has more emphasis on basic techniques that work under real-world conditions and less on more esoteric mathematics that has intrinsic elegance but less practical applicability.'' The mathematics covered is nicely presented, and there are applications, which may be used in traditional mathematics courses to illustrate the usefulness of, e.g., linear algebra, projective geometry, Fourier analysis etc. Each chapter contains exercises and references to additional reading. The first chapter gives a very nice introduction to the subject and then goes on to a historical overview of the development since 1970, as seen from the author's point of view. In this part, the author discusses in particular the philosophy which also lies behind this book on how to structure a task: 1) computational theory: what is the goal of the computation (task), and what are the known restraints; 2) representations and algorithms: representation of the input, output and intermediate information, and algorithms used; 3) hardware implementation (and point 3 will influence point 2). In Chapter 15, the conclusion, the author again gives another description of these three points and the philosophy of the book: ``I have attempted to instill in my students and in readers of this book a discipline founded on principles of engineering, science and statistics.'' Computer vision is a huge field, it draws from many areas of mathematics. The last section of Chapter 1 gives an overview of the book with the structure of a 10-week and a 13-week course, ideas for teaching (pose small programming projects rather than written homework or quizzes). We give some details for Chapters 2 and 3, and list the topics of the remaining chapters. Chapter 2, on image formation, is a very nice combination of mathematics and the relevant applications, keeping the level of mathematics to what is required. Among others, the following topics are addressed: 2D, 3D and subsets: lines, planes, conics; transformations: translations, rotations; projective geometry and homogeneous coordinates; representation of rotations with quaternions; concrete applications, e.g., from one camera to another; distortion; photometric image formation; reflectance and shading; the digital camera; signal analysis; colors. The topic of Chapter 3 is image processing. Among others, it deals with: filters (linear, non-linear); Fourier transforms; Wiener filtering; wavelets; geometric transformations; global optimization. As is typical for the book, many applications are given, namely to image blending, feature-based morphing and image restoration. Chapter 4: Feature detection and matching. Points and patches, edges, lines; Chapter 5: Segmentation; Chapter 6: Feature based alignment; Chapter 7: Structure from motion; Chapter 8: Dense motion estimation; Chapter 9: Image stitching; Chapter 10: Computational photography; Chapter 11: Stereo correspondence; Chapter 12: 3D reconstruction; Chapter 13: Image-based rendering; Chapter 14: Recognition; Chapter 15: Conclusion. Three appendices conclude the book, dealing with A) Linear algebra and numerical techniques; B) Bayesian modeling and inference; C) Supplementary material: Data sets, software, slides and lecture, bibliography. The book also contains many references to resources on the Internet.