Presumably this is an example of the emerging trend to shift the image processing programming paradigm over to Java. There is a tendency for the text itself to shift from instructional mode to reference manual mode slightly haphazardly.
Programming examples are quite detailed, so it is not suitable for novice programmers (whatever the language). Certainly you need to know Java to get any benefit from the first three chapters. Most of the usual friends from image processing are there, but the treatment of texturing and lighting is almost invisible. The maths is very light - deliberately it seems - which is either a blessing or a curse depending on needs.
There is little sustained emphasis in the text on object oriented design principles (doubtful if one would know anything significant about OO design having read the text)which is a definite weakness, but increasingly seems to dominate the roll out of Java texts. There is a chapter mid way through about rendering which develops a component class for image display which is very useful. To be fair the book doesn't set out to teach OO, but relies on using the Smalltalk Model-View-Controller pattern approach to program elaboration (cf. "Design Patterns" for a more thorough account).
As usual there is far too much code jammed into the text - a CD is better, giving more room on paper to tease out principles, design decisions and algorithms. However, the market doesn't agree it seems. The absence of in-depth discussion of algorithms (in an area replete with them) is a serious shortcoming. Chunks of code replace algorithm schema - pedagogical value of the text is limited therefore.
Overall, the book is uneven. There are pieces of work that are gems but a tighter focus on the interface between mathematical principles and algorithm generation would have produced a much better text (certainly for students).