Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Evolutionary computer vision : the first footprints
Olague G., Springer International Publishing, New York, NY, 2016. 411 pp. Type: Book (978-3-662436-92-9)
Date Reviewed: Jun 21 2017

Olague is not a newcomer to the fusion field of evolutionary computation and computer vision. He has published research articles on the nexus of these fields over the last two decades. Having personally taught courses on the individual fields, I find the combination a refreshing treatment. While the intent of the book is to cover advanced concepts, the author’s approach is to start from the very basics so that researchers from either field can immerse themselves in the new material. Each chapter is written as a self-contained article, starting with an abstract, introduction, history of the problem or concepts, mathematical or computational formulation, conclusions and future directions, and extensive references.

There are six parts to the book. Part 1 (“Introduction”) is a succinct description (8 pages) about the purpose and organization of the book. Part 2 (“Basics”) is a wonderful treatment of the fundamental concepts of computer vision, which sets the stage for incorporating evolutionary computations, covered in chapter 3. Chapter 3 on evolutionary computing is a classic treatment of the field, favoring a mathematical modeling approach.

In this part, the vision concepts follow David Marr’s philosophy [1] and the evolutionary concepts are presented from a pure Darwinian approach. Darwinian evolutionary theory is discussed from both general philosophical and focused teleological perspectives. There are various schools of thought about understanding vision and the role of evolutionary operators. This part was not intended to be an encyclopedic treatment of the subject matter, but a cohesive treatment of major approaches.

Part 3 (“Feature Location and Extraction”) consists of two chapters. Chapter 4 models corner features in general configuration (corner geometry) and specific types of junctions (such as K, L, T junctions, and so on). Again, this chapter is presented primarily as a global mathematical optimization problem, which the author argues will benefit greatly from an evolutionary computation approach. The results of the experimentation presented there suggest that he is correct in his assessment. Chapter 5 then develops interest point operators. The author builds a solid mathematical foundation, followed by an equal treatment of genetic programming operators (single and multiobjective/Pareto). The author balances stability, point dispersion, and information content.

Advanced concepts and applications are discussed in Part 4 (chapters 6 and 7) on 3D vision and multicriteria optimization. This part does not hesitate to use color figures and numerous diagrams and tables to elucidate the concepts. These chapters assume a fair knowledge of analytical geometry, linear algebra, and basic optimization techniques. In chapter 6, the author favors the cooperative honeybee search algorithm (based on exploration, recruitment, and harvest), which is a heuristic divide-and-conquer approach to optimization. To accomplish this, the author merges Dunn’s Parisian evolutionary computation approach [2] with Potter and De Jong’s cooperative evolutionary framework [3]. Chapter 7 accurately plans (sensor planning) photogrammetric networks.

Part 5 (“Learning and Recognition”) deals with a general approach to simultaneously solving region of interest (ROI) selection and the subsequent feature extraction tasks. The approach is based on a linear genetic programming algorithm where the members of the population are a sequence of instructions. Texture analysis is implemented by support vector machine (SVM) technology implemented by libSVM [4]. Finally, chapter 9 suggests a new approach for constructing feature descriptors that are region invariant. Invariance is the gold standard for computer vision operators because it guarantees that the features can be detected independent of where they occur in the image. The last part is a concise chapter that provides a summary and conclusions of the entire book and an appendix that provides the matrix mathematics of projections and camera calibrations.

The author has made a major contribution to anyone in the field of computer vision who is interested in alternative approaches to harder problems. This book also presents important applications to researchers in the evolutionary computation field, but that community may not be as familiar with the computer vision algorithms. Fortunately, the author provides the material you need to understand the prerequisite knowledge. To the general reader, it should be noted that although the treatments presented in this text are slightly more mathematical, the level of mathematics necessary is not beyond that which is studied in an undergraduate curriculum (calculus, linear algebra, and matrix optimization).

Reviewer:  R. Goldberg Review #: CR145364 (1708-0518)
1) Marr, D. Vision. W. H. Freeman & Co Ltd., Boston, MA, 1982.
2) Dunn, E.; Olague, G.; Lutton, E. Parisian camera placement for vision metrology. Pattern Recognition Letters 11, 8(2006), 1209–1219.
3) Potter, M.; De Jong, K. Cooperative coevolution: an architecture for evolving coadapted subcomponents. Evolutionary Computation 8, 1(2000), 1–29.
4) Chang, C. C.; Lin, C. J. libSVM, https://www.csie.ntu.edu.tw/~cjlin/libsvm/ (06/09/2017).
Bookmark and Share
  Reviewer Selected
Featured Reviewer
 
 
Vision And Scene Understanding (I.2.10 )
 
 
Applications (I.4.9 )
 
 
Scene Analysis (I.4.8 )
 
Would you recommend this review?
yes
no
Other reviews under "Vision And Scene Understanding": Date
Object recognition by computer
Grimson W., MIT Press, Cambridge, MA, 1990. Type: Book (9780262071307)
Mar 1 1992
Interaction with the environment
Coiffet P., Prentice-Hall, Inc., Upper Saddle River, NJ, 1983. Type: Book (9789780137821280)
Oct 1 1989
Vision: biology challenges technology
Ballard D., Brown C. BYTE 10(4): 245-261, 1985. Type: Article
Oct 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy