Browse Books

Reviewer: Morton Nadler

The first edition of this work [1] was nearly a comprehensive handbook of its subject matter. The second edition is not; to keep it to about the same size as the first edition, some topics that one would expect to find had to be omitted. Thus, we do not find tomographic reconstruction; image coding, especially image compression and decompression; or graphics (even though the author is with Sun after a distinguished career at the Image Processing Institute of the University of Southern California and several specialized companies). The author's “admission” that the book omits pattern recognition and image understanding (scene analysis) is superfluous. These topics are not properly the subject matter of digital image processing. So, what does the book cover__?__ Part 1, “Continuous Image Characterization,” presents mathematical and psychophysical image characterization, photometry, and colorimetry. Part 2, “Digital Image Characterization,” covers sampling and reconstruction, mathematical characterization of discrete images, and image quantization. Part 3, “Linear Processing,” discusses convolution, unitary transforms, and recursive filtering. Part 4, “Image Improvement,” describes enhancement, restoration, and geometrical transformation. Part<__?__Pub Fmt hardspace>5, “Image Analysis,” presents morphology, edge detection, feature extraction, segmentation, shape analysis, and detection and registration. All five parts are highly rigorous and <__?__Pub Fmt nolinebreak>mathematical.<__?__Pub Fmt /nolinebreak> Any book on image processing must be illustrated; this work is well illustrated. For those of us who have relied on the first edition, many of the old familiar figures are back in improved renderings, along with many new figures. Some of them are used as running examples, appearing over and over again to allow comparisons of differing techniques. Much of the bibliography has been retained from the first edition, and some recent material has been added. A curious self-referential twist appears on page 688, where the second edition is included in the bibliography. This excellent reference work should be on the shelves of every specialist in the field. I regret, however, that due to the self-imposed space limitations, this book, published in 1991, does not review image compression and especially image sequence compression. At least the bibliography could have referenced Netravali and Haskell's book [2]. On the other hand, Tanimoto's complaint in his review of the first edition<__?__Pub Fmt hardspace>[<__?__Pub Caret1>3] that “it does not, however, cover pattern classification” was unfair. Image processing is an entire discipline in its own right. It is a necessary auxiliary to visual pattern recognition, but to cover both subjects would have required a volume of twice the size or a two-volume set.