Digital Image Processing

Citation

Robinson, J.A. (2009), "Digital Image Processing", Sensor Review, Vol. 29 No. 4. https://doi.org/10.1108/sr.2009.08729dae.001

Digital Image Processing

Article Type: Book review From: Sensor Review, Volume 29, Issue 4

Rafael C. Gonzalez, Richard E. WoodsPrentice-HallAugust 21, 2007976 pp.ISBN: 9780131687288$155Web site: www.pearsonhighered.com/bookseller/academic/product/0,3110,013168728X,00.html

I encountered the first edition of Gonzalez and Wintz’s Digital Image Processing, published 1977, as a PhD student in the 1980s. It was becoming the textbook of choice for introductory courses and for lone readers like me, winning over chunkier and more demanding books from Pratt and Rosenfeld and Kak. Its accessibility and selective coverage made it ideal for the beginner. With a change of co-author in 1992, the coverage broadened and the page count grew. More practical, implementation-oriented books came out, and G&W found itself at the theoretical end of the market. Perhaps, influenced by the newer competition, the second Gonzalez and Woods edition in 2002 cut some of the math. But the third edition, out now, has put it back and expanded the text further. At 954 pages G&W is now one of the biggest image processing texts. Its long life shows it to be one of the most successful, but has it lost the clarity and focus of its younger days?

For this edition, the authors based their changes on a wide-ranging survey of readers throughout the world. The new material, they say, “is timely, highly readable, and illustrated with numerous examples”. Mostly, they are right. There are 400 new images – carefully chosen, illuminating and sometimes beautiful. New diagrams, new examples of analysis and application, help the serious reader understand and use concepts across a very broad field. For the first time there is coverage of tomography and refreshed chapters on compression and morphology. The readability is good, as claimed. The new material on point operators – luminance transformations and thresholding – is particularly welcome, and the revision of the math (particularly in dealing with the frequency domain and wavelets) is well-pitched and clear. A companion web site provides yet more material.

As the book has evolved, its chief virtue has become comprehensiveness. You cannot go wrong using G&W as a course text, provided you are able to guide students towards what really counts. Unfortunately for the lone reader, the authors offer little in the way of critical judgement. They rarely comment on the relative importance of the techniques they survey. They do not identify learning priorities or sections to skip. Certainly, it was time to include the Canny operator in the image segmentation chapter (perhaps three editions late), but are fuzzy sets important enough to image processing to justify their high page count compared to, say, smoothing and sharpening filters? The authors say that fuzzy set coverage was frequently requested in their survey. Perhaps, democracy is not the way to build a book! Meanwhile the compression chapter discusses threshold coding and zonal coding of transform coefficients. Later the chapter reviews Joint Photograph Experts Group (JPEG), but the authors offer no critique of how the scene-adaptive coding of coefficients adopted by JPEG effectively superseded both threshold and zonal coding 20 years ago!

G&W is great but includes too much. It may be time for a new textbook writer to turn out the new canonical textbook – only half the length of Digital Image Processing but with just what you need to be introduced to this exciting field.

John A. RobinsonDepartment of Electronics, University of York, York, UK