It took a long time, but the scientific communities finally agreed, tacitly, on a division of terminology. Vision generally refers to computational models in psychology; computer vision has to do with autonomous systems to mimic certain vision capabilities; image understanding is a code word hinting at technology support for vision problems in military applications; and pattern recognition has to do with discriminant function design for classification purposes. Thus it is slightly jarring that Pattern recognition by Mike James is mostly an undergraduate text on basic image processing.
The book is a monograph with many images and figures but no exercises, save for an appendix of 20 suggested projects. It comprises six chapters: 18 pages on notions of digital images, 17 pages on basic pattern recognition (linear classifiers, etc.), 28 pages on edge and line detection, 31 pages on basic Fourier theory for images, 18 pages on segmentation, and 32 pages on binary image processing and morphology. Some code is given in BASIC. Treatments throughout are standard, concise, and clear. For someone with no knowledge of computer vision or image processing, and only rudimentary knowledge of computer science, this book certainly provides a succinct introduction to the problems, issues, and techniques used in image processing. The book seems to fill a niche, whereas alternate texts such as Rosenfeld and Kak, Gonzalez and Wintz, or Pratt assume too much sophistication to assist one in sampling or entering the field.
On the other hand, given even these modest goals, it is not clear how well the book succeeds. There are still many formulas that would need considerable explanation for the uninitiated or novice student, and the presentation of algorithms in Microsoft QuickBASIC, with its use of percent signs to denote arrays, is a bit of a disaster. As a general principle, we should realize that if students or readers cannot convert clear pseudocode into proper compilable syntax for their favorite language on their available machine, then they are probably not going to follow the algorithm or concept anyway.
I would be reluctant to use this book as a text for an undergraduate course. It might serve as a supplemental text. To my mind, there is an obvious time when this book might be very useful. It is late in the term, the midterms still aren’t graded, there are eleven reviews to get done and a committee report to write, the annual conference paper should have been express-mailed yesterday, and a cherubic student or colleague from a different department comes into the office and says: “Say, I hear you do image processing. I’m interested in that. What is it?” Ah, I have over here a short book. . . .